{
  "_meta": {
    "document": "LLM Evolutionary Tree",
    "schemaVersion": "2026.05.02",
    "generatedAt": "2026-09-10T00:43:17.493Z",
    "canonicalUrl": "https://brianletort.ai/industry/tree",
    "yamlUrl": "https://brianletort.ai/industry/tree/models.yaml",
    "lastUpdated": "2026-09-05"
  },
  "branches": [
    {
      "id": "foundational",
      "name": "Foundational",
      "definition": "Pre-transformer language representations. Word embeddings and early contextual representations that established distributed semantics and the pretrain-then-finetune paradigm.\n",
      "branch_from": null
    },
    {
      "id": "encoder_only",
      "name": "Encoder-only",
      "definition": "Bidirectional transformer encoders trained with masked-language-modeling objectives for understanding tasks (classification, NER, QA, retrieval).\n",
      "branch_from": "transformer"
    },
    {
      "id": "encoder_decoder",
      "name": "Encoder-decoder",
      "definition": "Sequence-to-sequence transformers with separate encoder and decoder stacks; pretrained with denoising or text-to-text objectives.\n",
      "branch_from": "transformer"
    },
    {
      "id": "decoder_only",
      "name": "Decoder-only",
      "definition": "Autoregressive transformers that generate text token-by-token using only a decoder stack. The dominant architecture for chat models prior to the test-time-compute era.\n",
      "branch_from": "transformer"
    },
    {
      "id": "mixture_of_experts",
      "name": "Mixture-of-Experts",
      "definition": "Sparse architectures that route each token to a small subset of expert sub-networks. Lower per-token compute, higher total parameter count, and meaningfully different infrastructure profile (interconnect-heavy).\n",
      "branch_from": "transformer"
    },
    {
      "id": "multimodal",
      "name": "Multimodal",
      "definition": "Models trained natively on text plus image, audio, or video. The defining shift is unified representation across modalities, not a vision adapter bolted onto a text-only model.\n",
      "branch_from": "transformer"
    },
    {
      "id": "reasoning",
      "name": "Reasoning",
      "definition": "Models whose defining architectural shift is test-time compute — large-scale RL on chain-of-thought to produce long internal reasoning before a final answer. Often called \"thinking models\" or \"extended-thinking models.\"\n",
      "branch_from": "transformer"
    }
  ],
  "breakthroughs": [
    {
      "id": "seq2seq-attention",
      "name": "Seq2Seq with Attention",
      "year": 2014,
      "body": "Sutskever, Vinyals, and Le introduced the LSTM encoder-decoder for machine translation; Bahdanau, Cho, and Bengio added soft attention to fix the fixed-length context bottleneck. Together these papers are the direct ancestors of the transformer.\n",
      "affected_branches": [
        "foundational",
        "encoder_decoder"
      ]
    },
    {
      "id": "transformer",
      "name": "Transformer (Attention Is All You Need)",
      "year": 2017,
      "body": "Vaswani et al. replaced recurrence and convolution with self-attention alone. Achieved 28.4 BLEU on WMT'14 EN-DE in 3.5 days on 8 GPUs and became the shared ancestor of every modern LLM branch.\n",
      "affected_branches": [
        "encoder_only",
        "encoder_decoder",
        "decoder_only",
        "mixture_of_experts",
        "multimodal",
        "reasoning"
      ]
    },
    {
      "id": "mlm-pretraining",
      "name": "Masked Language Modeling Pretraining",
      "year": 2018,
      "body": "BERT (Devlin et al.) established bidirectional masked-language-modeling pretraining as the foundation of modern NLP understanding. Defined the encoder-only branch.\n",
      "affected_branches": [
        "encoder_only"
      ]
    },
    {
      "id": "text-to-text-framing",
      "name": "Text-to-Text Framing",
      "year": 2019,
      "body": "T5 (Raffel et al.) reframed every NLP task as text-to-text and established span-corruption denoising on the C4 corpus. Defined the modern encoder-decoder branch.\n",
      "affected_branches": [
        "encoder_decoder"
      ]
    },
    {
      "id": "in-context-learning",
      "name": "In-Context Learning at Scale",
      "year": 2020,
      "body": "GPT-3 (Brown et al.) demonstrated few-shot in-context learning as an emergent capability of scale. The 175B-parameter dense decoder became the scaling-laws moment and launched the OpenAI API.\n",
      "affected_branches": [
        "decoder_only"
      ]
    },
    {
      "id": "rlhf-instructgpt",
      "name": "RLHF / InstructGPT",
      "year": 2022,
      "body": "Reinforcement Learning from Human Feedback applied to GPT-3 produced InstructGPT and then ChatGPT. Made dialogue-format chat tuning the industry default and turned LLMs into products.\n",
      "affected_branches": [
        "decoder_only"
      ]
    },
    {
      "id": "chinchilla-scaling",
      "name": "Chinchilla Compute-Optimal Scaling",
      "year": 2022,
      "body": "Hoffmann et al. (DeepMind) established that tokens and parameters should scale equally for compute-optimal training. Outperformed Gopher 280B and GPT-3 175B on MMLU at 70B parameters and 1.3T tokens. Every subsequent training run respects this law.\n",
      "affected_branches": [
        "decoder_only",
        "mixture_of_experts",
        "multimodal",
        "reasoning"
      ]
    },
    {
      "id": "switch-transformer-moe",
      "name": "Trillion-Parameter Sparse Routing",
      "year": 2021,
      "body": "Switch Transformer (Fedus, Zoph, Shazeer) pushed mixture-of-experts past one trillion parameters with simplified single-expert routing. 4-7x speedup over T5-XXL. Defined the MoE branch at scale.\n",
      "affected_branches": [
        "mixture_of_experts"
      ]
    },
    {
      "id": "flash-attention",
      "name": "FlashAttention",
      "year": 2022,
      "body": "Dao et al. produced an IO-aware exact attention algorithm that reduced the memory footprint of attention from quadratic to linear in sequence length. Made long-context training and inference economically feasible at scale.\n",
      "affected_branches": [
        "decoder_only",
        "mixture_of_experts",
        "multimodal",
        "reasoning"
      ]
    },
    {
      "id": "native-multimodal",
      "name": "Native Multimodal Training",
      "year": 2022,
      "body": "Flamingo (Alayrac et al., DeepMind) demonstrated few-shot vision-language learning by interleaving images, video, and text in a single sequence. Defined the multimodal branch and seeded Gemini's natively-multimodal design.\n",
      "affected_branches": [
        "multimodal"
      ]
    },
    {
      "id": "moe-go-mainstream",
      "name": "Open MoE Goes Mainstream",
      "year": 2023,
      "body": "Mixtral 8x7B (Dec 2023) was the first widely-deployed open-weights mixture-of-experts. Apache 2.0 license; 47B total / 13B active. Triggered an open-source MoE wave (DBRX, Arctic, Jamba, DeepSeek-V2).\n",
      "affected_branches": [
        "mixture_of_experts"
      ]
    },
    {
      "id": "multi-head-latent-attention",
      "name": "Multi-head Latent Attention (MLA)",
      "year": 2024,
      "body": "DeepSeek-V2 introduced MLA — a low-rank latent compression of the KV cache that reduced memory by 93% versus dense attention. Combined with DeepSeekMoE, this is the architectural innovation that enabled the V3/R1 cost shock in early 2025.\n",
      "affected_branches": [
        "mixture_of_experts",
        "reasoning"
      ]
    },
    {
      "id": "test-time-compute",
      "name": "Test-Time Compute (the o1 moment)",
      "year": 2024,
      "body": "OpenAI's o1-preview (September 2024) demonstrated that large-scale RL on chain-of-thought, with the model \"thinking\" before answering, produces capability gains comparable to a generation of pretraining scale-up. Defined the reasoning branch.\n",
      "affected_branches": [
        "reasoning"
      ]
    },
    {
      "id": "grpo-rl",
      "name": "Group Relative Policy Optimization (GRPO)",
      "year": 2025,
      "body": "DeepSeek introduced GRPO in DeepSeek-Math and scaled it in DeepSeek-R1 to skip supervised fine-tuning entirely and train reasoning behavior via pure RL. Lower training cost than RLHF/PPO and competitive results. Now used by Qwen, Microsoft Phi reasoning, Hunyuan, GLM, and others.\n",
      "affected_branches": [
        "reasoning"
      ]
    },
    {
      "id": "mcp",
      "name": "Model Context Protocol (MCP)",
      "year": 2024,
      "body": "Anthropic published the Model Context Protocol — an open spec for how LLMs connect to tools, data sources, and other models. Adopted by every major frontier vendor by mid-2025. Made the agentic stack interoperable.\n",
      "affected_branches": [
        "decoder_only",
        "multimodal",
        "reasoning"
      ]
    },
    {
      "id": "a2a",
      "name": "Agent-to-Agent (A2A) Handoff",
      "year": 2025,
      "body": "A2A protocol (Google + Anthropic + OpenAI cross-vendor work) standardized structured handoffs between agents. Enabled native multi-agent orchestration in Kimi K2.6, Grok 4.20, GLM-5.1.\n",
      "affected_branches": [
        "reasoning"
      ]
    },
    {
      "id": "deepseek-sparse-attention",
      "name": "DeepSeek Sparse Attention (DSA)",
      "year": 2025,
      "body": "DeepSeek-V3.2 introduced Sparse Attention — a per-head dynamic sparsity pattern that dropped attention compute by ~73% per token. First cross-vendor architectural pattern to propagate within the Chinese frontier itself; adopted by GLM-5 (Feb 2026) and reused in DeepSeek-V4.\n",
      "affected_branches": [
        "mixture_of_experts",
        "reasoning"
      ]
    },
    {
      "id": "adaptive-thinking",
      "name": "Adaptive Thinking",
      "year": 2026,
      "body": "Claude Opus 4.6 (Feb 2026) introduced adaptive thinking — the model auto-decides reasoning depth per query rather than requiring an explicit \"thinking mode\" toggle. Replaces manual extended thinking. Now used by Mistral Small 4, Qwen3.6, Doubao 1.6.\n",
      "affected_branches": [
        "reasoning"
      ]
    }
  ],
  "models": [
    {
      "id": "word2vec",
      "name": "Word2Vec",
      "vendor": "Google",
      "release_date": "2013-01-16",
      "branch": "foundational",
      "parents": [],
      "influences": [],
      "openness": "open_source",
      "parameters": "n/a",
      "context": "n/a",
      "modalities": [
        "text"
      ],
      "capabilities": [],
      "secondary_tags": [
        "open_source",
        "foundational",
        "embedding"
      ],
      "notable": "Established distributed word representations as the foundation of modern NLP. Trained 300-dim vectors over 1.6B words in under 1 day on a single CPU.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "glove",
      "name": "GloVe",
      "vendor": "Stanford NLP",
      "release_date": "2014-10-25",
      "branch": "foundational",
      "parents": [
        "word2vec"
      ],
      "influences": [
        "word2vec"
      ],
      "openness": "open_source",
      "parameters": "n/a",
      "context": "n/a",
      "modalities": [
        "text"
      ],
      "capabilities": [],
      "secondary_tags": [
        "open_source",
        "foundational",
        "embedding"
      ],
      "notable": "75% accuracy on word analogy tasks at release; ~30K citations. Pennington/Socher/Manning, EMNLP 2014.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "fasttext",
      "name": "FastText",
      "vendor": "Meta",
      "release_date": "2016-07-15",
      "branch": "foundational",
      "parents": [
        "word2vec"
      ],
      "influences": [
        "word2vec"
      ],
      "openness": "open_source",
      "parameters": "n/a",
      "context": "n/a",
      "modalities": [
        "text"
      ],
      "capabilities": [],
      "secondary_tags": [
        "open_source",
        "foundational",
        "embedding",
        "multilingual"
      ],
      "notable": "Mikolov's second-act embedding work at Facebook. Introduced subword information; foundation of Facebook's open-source fastText library.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "elmo",
      "name": "ELMo",
      "vendor": "Allen Institute for AI",
      "release_date": "2018-02-15",
      "branch": "foundational",
      "parents": [
        "word2vec",
        "glove"
      ],
      "influences": [],
      "openness": "open_source",
      "parameters": "~94M (biLM)",
      "context": "n/a",
      "modalities": [
        "text"
      ],
      "capabilities": [],
      "secondary_tags": [
        "open_source",
        "foundational"
      ],
      "notable": "First widely-adopted contextualized word representation. NAACL 2018 best paper. Last major pre-transformer milestone before BERT.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "ulmfit",
      "name": "ULMFiT",
      "vendor": "fast.ai",
      "release_date": "2018-01-18",
      "branch": "foundational",
      "parents": [],
      "influences": [],
      "openness": "open_source",
      "parameters": "~24M (AWD-LSTM)",
      "context": "n/a",
      "modalities": [
        "text"
      ],
      "capabilities": [],
      "secondary_tags": [
        "open_source",
        "foundational"
      ],
      "notable": "Established the pretrain-then-finetune paradigm for NLP that GPT and BERT generalized later in 2018. Howard and Ruder, ACL 2018.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "bert-base",
      "name": "BERT (base)",
      "vendor": "Google",
      "release_date": "2018-10-11",
      "branch": "encoder_only",
      "parents": [],
      "influences": [],
      "openness": "open_source",
      "parameters": "110M",
      "context": "512",
      "modalities": [
        "text"
      ],
      "capabilities": [],
      "secondary_tags": [
        "open_source",
        "foundational"
      ],
      "notable": "Defined the encoder-only branch; first widely-deployed bidirectional MLM pretraining objective.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "bert-large",
      "name": "BERT (large)",
      "vendor": "Google",
      "release_date": "2018-10-11",
      "branch": "encoder_only",
      "parents": [
        "bert-base"
      ],
      "influences": [],
      "openness": "open_source",
      "parameters": "340M",
      "context": "512",
      "modalities": [
        "text"
      ],
      "capabilities": [],
      "secondary_tags": [
        "open_source",
        "foundational"
      ],
      "notable": "Topped GLUE and SQuAD on release.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "roberta",
      "name": "RoBERTa",
      "vendor": "Meta",
      "release_date": "2019-07-26",
      "branch": "encoder_only",
      "parents": [
        "bert-base"
      ],
      "influences": [],
      "openness": "open_source",
      "parameters": "125M (base) / 355M (large)",
      "context": "512",
      "modalities": [
        "text"
      ],
      "capabilities": [],
      "secondary_tags": [
        "open_source"
      ],
      "notable": "Showed BERT was significantly under-trained.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "albert",
      "name": "ALBERT",
      "vendor": "Google",
      "release_date": "2019-09-26",
      "branch": "encoder_only",
      "parents": [
        "bert-base"
      ],
      "influences": [],
      "openness": "open_source",
      "parameters": "12M / 18M / 60M / 235M (xxlarge)",
      "context": "512",
      "modalities": [
        "text"
      ],
      "capabilities": [],
      "secondary_tags": [
        "open_source",
        "small"
      ],
      "notable": "Lite BERT; xxlarge has fewer parameters than BERT-large but outperforms it on GLUE/SQuAD/RACE.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "distilbert",
      "name": "DistilBERT",
      "vendor": "HuggingFace",
      "release_date": "2019-10-02",
      "branch": "encoder_only",
      "parents": [
        "bert-base"
      ],
      "influences": [],
      "openness": "open_source",
      "parameters": "66M",
      "context": "512",
      "modalities": [
        "text"
      ],
      "capabilities": [],
      "secondary_tags": [
        "open_source",
        "distilled",
        "small",
        "edge"
      ],
      "notable": "40% smaller, 60% faster, retains ~97% of BERT's GLUE performance. Foundational for on-device NLP.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "xlnet",
      "name": "XLNet",
      "vendor": "CMU + Google",
      "release_date": "2019-06-19",
      "branch": "encoder_only",
      "parents": [
        "bert-base"
      ],
      "influences": [],
      "openness": "open_source",
      "parameters": "110M (base) / 340M (large)",
      "context": "512",
      "modalities": [
        "text"
      ],
      "capabilities": [],
      "secondary_tags": [
        "open_source"
      ],
      "notable": "First major bridge between AR (GPT-style) and bidirectional (BERT-style) pretraining.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "electra",
      "name": "ELECTRA",
      "vendor": "Google",
      "release_date": "2020-03-23",
      "branch": "encoder_only",
      "parents": [
        "bert-base"
      ],
      "influences": [],
      "openness": "open_source",
      "parameters": "14M (small) / 110M (base) / 335M (large)",
      "context": "512",
      "modalities": [
        "text"
      ],
      "capabilities": [],
      "secondary_tags": [
        "open_source",
        "small"
      ],
      "notable": "Matches RoBERTa/XLNet quality at a fraction of compute; ELECTRA-small runs on a single GPU.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "ernie-3",
      "name": "ERNIE 3.0",
      "vendor": "Baidu",
      "release_date": "2021-07-05",
      "branch": "encoder_only",
      "parents": [],
      "influences": [
        "bert-base",
        "t5"
      ],
      "openness": "open_source",
      "parameters": "10B (ERNIE 3.0); 260B (ERNIE 3.0 Titan)",
      "context": "512",
      "modalities": [
        "text"
      ],
      "capabilities": [],
      "secondary_tags": [
        "open_source",
        "multilingual",
        "china"
      ],
      "notable": "Topped SuperGLUE (90.6%) — first to exceed human baseline on July 3, 2021. Titan was the largest Chinese dense LM at the time.\n",
      "status": "active",
      "placement_confidence": "medium"
    },
    {
      "id": "text-embedding-3",
      "name": "text-embedding-3",
      "vendor": "OpenAI",
      "release_date": "2024-01-25",
      "branch": "encoder_only",
      "parents": [],
      "influences": [
        "gpt-3"
      ],
      "openness": "closed",
      "parameters": "undisclosed",
      "context": "8192",
      "modalities": [
        "text"
      ],
      "capabilities": [],
      "secondary_tags": [
        "closed",
        "embedding"
      ],
      "notable": "OpenAI's only major encoder-only entry; included for architectural completeness.\n",
      "status": "active",
      "placement_confidence": "medium"
    },
    {
      "id": "t5",
      "name": "T5",
      "vendor": "Google",
      "release_date": "2019-10-23",
      "branch": "encoder_decoder",
      "parents": [],
      "influences": [],
      "openness": "open_source",
      "parameters": "60M / 220M / 770M / 3B / 11B",
      "context": "512",
      "modalities": [
        "text"
      ],
      "capabilities": [],
      "secondary_tags": [
        "open_source",
        "foundational"
      ],
      "notable": "Defined the modern encoder-decoder line; every NLP task framed as text-to-text. Anchored Switch Transformer, UL2, FLAN-T5.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "bart",
      "name": "BART",
      "vendor": "Meta",
      "release_date": "2019-10-29",
      "branch": "encoder_decoder",
      "parents": [],
      "influences": [
        "bert-base"
      ],
      "openness": "open_source",
      "parameters": "140M (base) / 400M (large)",
      "context": "1024",
      "modalities": [
        "text"
      ],
      "capabilities": [],
      "secondary_tags": [
        "open_source"
      ],
      "notable": "Combines BERT (bidirectional encoder) and GPT (autoregressive decoder). Workhorse for summarization.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "t0",
      "name": "T0",
      "vendor": "BigScience",
      "release_date": "2021-10-15",
      "branch": "encoder_decoder",
      "parents": [
        "t5"
      ],
      "influences": [],
      "openness": "open_source",
      "parameters": "3B / 11B",
      "context": "1024",
      "modalities": [
        "text"
      ],
      "capabilities": [],
      "secondary_tags": [
        "open_source",
        "multilingual"
      ],
      "notable": "Outperformed GPT-3 (175B) zero-shot on many tasks while being 16x smaller.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "ul2",
      "name": "UL2",
      "vendor": "Google",
      "release_date": "2022-05-10",
      "branch": "encoder_decoder",
      "parents": [
        "t5"
      ],
      "influences": [],
      "openness": "open_source",
      "parameters": "20B",
      "context": "2048",
      "modalities": [
        "text"
      ],
      "capabilities": [],
      "secondary_tags": [
        "open_source"
      ],
      "notable": "Beat GPT-3 zero-shot SuperGLUE; later instruction-tuned as Flan-UL2.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "glm",
      "name": "GLM (original)",
      "vendor": "Tsinghua THUDM",
      "release_date": "2021-03-18",
      "branch": "encoder_decoder",
      "parents": [],
      "influences": [],
      "openness": "open_source",
      "parameters": "110M to 130B (GLM-130B)",
      "context": "2K",
      "modalities": [
        "text"
      ],
      "capabilities": [],
      "secondary_tags": [
        "open_source",
        "bilingual",
        "china"
      ],
      "notable": "Bilingual (English plus Chinese); GLM-130B (October 2022) outperformed GPT-3 175B and BLOOM-176B on LAMBADA/MMLU. Anchored ChatGLM/GLM-4.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "flan-t5",
      "name": "FLAN-T5",
      "vendor": "Google",
      "release_date": "2022-10-20",
      "branch": "encoder_decoder",
      "parents": [
        "t5"
      ],
      "influences": [
        "t0"
      ],
      "openness": "open_source",
      "parameters": "80M / 250M / 780M / 3B / 11B",
      "context": "1024",
      "modalities": [
        "text"
      ],
      "capabilities": [],
      "secondary_tags": [
        "open_source"
      ],
      "notable": "Reference instruction-tuned encoder-decoder. Showed instruction tuning plus CoT scaling generalizes.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "gpt-1",
      "name": "GPT-1",
      "vendor": "OpenAI",
      "release_date": "2018-06-11",
      "branch": "decoder_only",
      "parents": [],
      "influences": [],
      "openness": "closed",
      "parameters": "117M",
      "context": "512",
      "modalities": [
        "text"
      ],
      "capabilities": [],
      "secondary_tags": [
        "closed",
        "foundational"
      ],
      "notable": "Origin of the GPT family. Shipped as the \"Improving Language Understanding\" research paper, not a product.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "gpt-2",
      "name": "GPT-2",
      "vendor": "OpenAI",
      "release_date": "2019-02-14",
      "branch": "decoder_only",
      "parents": [
        "gpt-1"
      ],
      "influences": [],
      "openness": "open_weights",
      "parameters": "1.5B",
      "context": "1024",
      "modalities": [
        "text"
      ],
      "capabilities": [
        "open_weights"
      ],
      "secondary_tags": [
        "open_weights",
        "historical_frontier"
      ],
      "notable": "First major \"too dangerous to release\" moment in modern LLMs. The staged release set the template for capability/safety tradeoffs.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "gpt-3",
      "name": "GPT-3",
      "vendor": "OpenAI",
      "release_date": "2020-06-11",
      "branch": "decoder_only",
      "parents": [
        "gpt-2"
      ],
      "influences": [],
      "openness": "closed",
      "parameters": "175B",
      "context": "2048",
      "modalities": [
        "text"
      ],
      "capabilities": [],
      "secondary_tags": [
        "closed",
        "historical_frontier"
      ],
      "notable": "The scaling-laws moment. First closed-API LLM at this size; launched the OpenAI API.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "gpt-3-5-turbo",
      "name": "GPT-3.5 Turbo",
      "vendor": "OpenAI",
      "release_date": "2022-11-30",
      "branch": "decoder_only",
      "parents": [
        "gpt-3"
      ],
      "influences": [],
      "openness": "closed",
      "parameters": "~20B (estimated; undisclosed)",
      "context": "4K (later 16K)",
      "modalities": [
        "text"
      ],
      "capabilities": [
        "tools"
      ],
      "secondary_tags": [
        "closed",
        "historical_frontier",
        "tool_using"
      ],
      "notable": "Launched ChatGPT (Nov 30, 2022), the fastest consumer software to 100M users. Made RLHF chat tuning the industry default.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "gpt-4",
      "name": "GPT-4",
      "vendor": "OpenAI",
      "release_date": "2023-03-14",
      "branch": "decoder_only",
      "parents": [
        "gpt-3-5-turbo"
      ],
      "influences": [],
      "openness": "closed",
      "parameters": "undisclosed",
      "context": "8K (later 32K variants)",
      "modalities": [
        "text"
      ],
      "capabilities": [
        "tools"
      ],
      "secondary_tags": [
        "closed",
        "historical_frontier",
        "tool_using"
      ],
      "notable": "Set the bar for closed-frontier models for ~14 months. Retired from ChatGPT April 30, 2025.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "gpt-4-turbo",
      "name": "GPT-4 Turbo",
      "vendor": "OpenAI",
      "release_date": "2023-11-06",
      "branch": "decoder_only",
      "parents": [
        "gpt-4"
      ],
      "influences": [],
      "openness": "closed",
      "parameters": "undisclosed",
      "context": "128K",
      "modalities": [
        "text",
        "image"
      ],
      "capabilities": [
        "tools",
        "multimodal",
        "long_context"
      ],
      "secondary_tags": [
        "closed",
        "long_context",
        "tool_using"
      ],
      "notable": "Established 128K as the de-facto context-window expectation for frontier models.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "gpt-4-5",
      "name": "GPT-4.5",
      "vendor": "OpenAI",
      "release_date": "2025-02-27",
      "branch": "decoder_only",
      "parents": [
        "gpt-4-turbo"
      ],
      "influences": [
        "gpt-4o"
      ],
      "openness": "closed",
      "parameters": "undisclosed",
      "context": "128K",
      "modalities": [
        "text",
        "image"
      ],
      "capabilities": [
        "tools",
        "multimodal",
        "long_context"
      ],
      "secondary_tags": [
        "closed",
        "historical_frontier",
        "tool_using"
      ],
      "notable": "Marked the end of pure-pretraining-scaling as OpenAI's frontier strategy. The next flagship (GPT-5) was unified with reasoning.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "claude-1",
      "name": "Claude 1",
      "vendor": "Anthropic",
      "release_date": "2023-03-14",
      "branch": "decoder_only",
      "parents": [],
      "influences": [],
      "openness": "closed",
      "parameters": "undisclosed",
      "context": "9K, expanded to 100K (May 2023)",
      "modalities": [
        "text"
      ],
      "capabilities": [
        "long_context"
      ],
      "secondary_tags": [
        "closed",
        "historical_frontier",
        "long_context"
      ],
      "notable": "Anthropic's first public model. May 2023 100K context window was an industry first.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "claude-instant-1",
      "name": "Claude Instant 1",
      "vendor": "Anthropic",
      "release_date": "2023-03-14",
      "branch": "decoder_only",
      "parents": [
        "claude-1"
      ],
      "influences": [],
      "openness": "closed",
      "parameters": "undisclosed (smaller than Claude 1)",
      "context": "100K",
      "modalities": [
        "text"
      ],
      "capabilities": [
        "long_context"
      ],
      "secondary_tags": [
        "closed",
        "small",
        "long_context"
      ],
      "notable": "Anthropic's first cost/latency tier; direct ancestor of the Haiku branch.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "claude-2",
      "name": "Claude 2",
      "vendor": "Anthropic",
      "release_date": "2023-07-11",
      "branch": "decoder_only",
      "parents": [
        "claude-1"
      ],
      "influences": [],
      "openness": "closed",
      "parameters": "undisclosed",
      "context": "100K",
      "modalities": [
        "text"
      ],
      "capabilities": [
        "long_context"
      ],
      "secondary_tags": [
        "closed",
        "long_context"
      ],
      "notable": "First Claude broadly available via API; coding plus math gains over Claude 1.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "claude-2-1",
      "name": "Claude 2.1",
      "vendor": "Anthropic",
      "release_date": "2023-11-21",
      "branch": "decoder_only",
      "parents": [
        "claude-2"
      ],
      "influences": [],
      "openness": "closed",
      "parameters": "undisclosed",
      "context": "200K",
      "modalities": [
        "text"
      ],
      "capabilities": [
        "tools",
        "long_context"
      ],
      "secondary_tags": [
        "closed",
        "long_context",
        "tool_using"
      ],
      "notable": "First Claude with 200K context; first tool-use beta. Retired July 21, 2025.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "claude-3-haiku",
      "name": "Claude 3 Haiku",
      "vendor": "Anthropic",
      "release_date": "2024-03-13",
      "branch": "decoder_only",
      "parents": [
        "claude-2-1"
      ],
      "influences": [
        "claude-instant-1"
      ],
      "openness": "closed",
      "parameters": "undisclosed",
      "context": "200K",
      "modalities": [
        "text",
        "image"
      ],
      "capabilities": [
        "tools",
        "multimodal",
        "long_context"
      ],
      "secondary_tags": [
        "closed",
        "small",
        "long_context",
        "multimodal"
      ],
      "notable": "First multimodal Claude tier.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "claude-3-sonnet",
      "name": "Claude 3 Sonnet",
      "vendor": "Anthropic",
      "release_date": "2024-03-04",
      "branch": "decoder_only",
      "parents": [
        "claude-2-1"
      ],
      "influences": [],
      "openness": "closed",
      "parameters": "undisclosed",
      "context": "200K",
      "modalities": [
        "text",
        "image"
      ],
      "capabilities": [
        "tools",
        "multimodal",
        "long_context"
      ],
      "secondary_tags": [
        "closed",
        "long_context",
        "multimodal"
      ],
      "notable": "Balanced cost/intelligence tier of Claude 3.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "claude-3-opus",
      "name": "Claude 3 Opus",
      "vendor": "Anthropic",
      "release_date": "2024-03-04",
      "branch": "decoder_only",
      "parents": [
        "claude-2-1"
      ],
      "influences": [],
      "openness": "closed",
      "parameters": "undisclosed",
      "context": "200K",
      "modalities": [
        "text",
        "image"
      ],
      "capabilities": [
        "tools",
        "multimodal",
        "long_context"
      ],
      "secondary_tags": [
        "closed",
        "historical_frontier",
        "long_context",
        "multimodal"
      ],
      "notable": "Anthropic's first model to surpass GPT-4 on multiple benchmarks.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "claude-3-5-sonnet",
      "name": "Claude 3.5 Sonnet",
      "vendor": "Anthropic",
      "release_date": "2024-06-20",
      "branch": "decoder_only",
      "parents": [
        "claude-3-sonnet"
      ],
      "influences": [],
      "openness": "closed",
      "parameters": "undisclosed",
      "context": "200K",
      "modalities": [
        "text",
        "image"
      ],
      "capabilities": [
        "tools",
        "multimodal",
        "long_context",
        "computer_use"
      ],
      "secondary_tags": [
        "closed",
        "long_context",
        "multimodal",
        "agentic",
        "computer_use"
      ],
      "notable": "Introduced computer-use API in October 2024 — first agentic GUI-control SKU.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "claude-3-5-haiku",
      "name": "Claude 3.5 Haiku",
      "vendor": "Anthropic",
      "release_date": "2024-11-04",
      "branch": "decoder_only",
      "parents": [
        "claude-3-haiku"
      ],
      "influences": [],
      "openness": "closed",
      "parameters": "undisclosed",
      "context": "200K",
      "modalities": [
        "text",
        "image"
      ],
      "capabilities": [
        "tools",
        "long_context",
        "multimodal"
      ],
      "secondary_tags": [
        "closed",
        "small",
        "long_context",
        "tool_using",
        "multimodal"
      ],
      "notable": "Matched Claude 3 Opus performance at Haiku-tier cost.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "lamda",
      "name": "LaMDA",
      "vendor": "Google",
      "release_date": "2021-05-18",
      "branch": "decoder_only",
      "parents": [],
      "influences": [],
      "openness": "closed",
      "parameters": "Up to 137B",
      "context": "Not disclosed",
      "modalities": [
        "text"
      ],
      "capabilities": [],
      "secondary_tags": [
        "closed"
      ],
      "notable": "Powered the original Bard at March 2023 launch.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "palm",
      "name": "PaLM",
      "vendor": "Google",
      "release_date": "2022-04-04",
      "branch": "decoder_only",
      "parents": [],
      "influences": [],
      "openness": "closed",
      "parameters": "8B / 62B / 540B",
      "context": "2048",
      "modalities": [
        "text"
      ],
      "capabilities": [],
      "secondary_tags": [
        "closed",
        "historical_frontier"
      ],
      "notable": "Largest dense LM at release; chain-of-thought reasoning emerged at 540B scale.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "palm-2",
      "name": "PaLM 2",
      "vendor": "Google",
      "release_date": "2023-05-10",
      "branch": "decoder_only",
      "parents": [
        "palm"
      ],
      "influences": [
        "chinchilla"
      ],
      "openness": "closed",
      "parameters": "Undisclosed; sizes Gecko / Otter / Bison / Unicorn",
      "context": "8K",
      "modalities": [
        "text"
      ],
      "capabilities": [],
      "secondary_tags": [
        "closed",
        "multilingual"
      ],
      "notable": "Powered Bard from May 2023 to Dec 2023 transition.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "chinchilla",
      "name": "Chinchilla",
      "vendor": "DeepMind",
      "release_date": "2022-03-29",
      "branch": "decoder_only",
      "parents": [],
      "influences": [],
      "openness": "research_only",
      "parameters": "70B (trained on 1.3T tokens)",
      "context": "2048",
      "modalities": [
        "text"
      ],
      "capabilities": [],
      "secondary_tags": [
        "research_only",
        "foundational"
      ],
      "notable": "Established compute-optimal training law: tokens and parameters should scale equally. Outperformed Gopher (280B) and GPT-3 (175B) on MMLU.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "sparrow",
      "name": "Sparrow",
      "vendor": "DeepMind",
      "release_date": "2022-09-28",
      "branch": "decoder_only",
      "parents": [
        "chinchilla"
      ],
      "influences": [],
      "openness": "research_only",
      "parameters": "70B (Chinchilla backbone, fine-tuned)",
      "context": "2048",
      "modalities": [
        "text"
      ],
      "capabilities": [],
      "secondary_tags": [
        "research_only"
      ],
      "notable": "Research-only; never productized. Influenced safety methodology in Bard / Gemini and broader RLHF practice.\n",
      "status": "research_only",
      "placement_confidence": "high"
    },
    {
      "id": "gemma-1",
      "name": "Gemma 1",
      "vendor": "Google",
      "release_date": "2024-02-21",
      "branch": "decoder_only",
      "parents": [],
      "influences": [
        "gemini-1"
      ],
      "openness": "open_weights",
      "parameters": "2B / 7B",
      "context": "8K",
      "modalities": [
        "text"
      ],
      "capabilities": [
        "open_weights"
      ],
      "secondary_tags": [
        "open_weights",
        "small",
        "edge"
      ],
      "notable": "Google's open-weights re-entry.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "gemma-2",
      "name": "Gemma 2",
      "vendor": "Google",
      "release_date": "2024-06-27",
      "branch": "decoder_only",
      "parents": [
        "gemma-1"
      ],
      "influences": [
        "gemini-1-5"
      ],
      "openness": "open_weights",
      "parameters": "2B / 9B / 27B",
      "context": "8K",
      "modalities": [
        "text"
      ],
      "capabilities": [
        "open_weights"
      ],
      "secondary_tags": [
        "open_weights",
        "small"
      ],
      "notable": "27B variant matched Llama-3-70B on many benchmarks at ~1/3 the parameters.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "llama-1",
      "name": "LLaMA 1",
      "vendor": "Meta AI",
      "release_date": "2023-02-24",
      "branch": "decoder_only",
      "parents": [],
      "influences": [
        "chinchilla",
        "palm"
      ],
      "openness": "research_only",
      "parameters": "7B, 13B, 33B, 65B",
      "context": "2K",
      "modalities": [
        "text"
      ],
      "capabilities": [],
      "secondary_tags": [
        "research_only",
        "foundational"
      ],
      "notable": "Set the recipe (RMSNorm/SwiGLU/RoPE) every Western open model now follows.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "llama-2",
      "name": "Llama 2",
      "vendor": "Meta AI",
      "release_date": "2023-07-18",
      "branch": "decoder_only",
      "parents": [
        "llama-1"
      ],
      "influences": [],
      "openness": "open_weights",
      "parameters": "7B, 13B, 70B",
      "context": "4K",
      "modalities": [
        "text"
      ],
      "capabilities": [
        "open_weights"
      ],
      "secondary_tags": [
        "open_weights"
      ],
      "notable": "First commercially-licensed Llama; legitimized open-weights ecosystem outside research.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "code-llama",
      "name": "Code Llama",
      "vendor": "Meta AI",
      "release_date": "2023-08-24",
      "branch": "decoder_only",
      "parents": [
        "llama-2"
      ],
      "influences": [],
      "openness": "open_weights",
      "parameters": "7B, 13B, 34B, 70B",
      "context": "100K (effective)",
      "modalities": [
        "text"
      ],
      "capabilities": [
        "open_weights"
      ],
      "secondary_tags": [
        "open_weights",
        "code_specialist"
      ],
      "notable": "Anchor of the open-coder branch; precursor to most open coding fine-tunes.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "llama-3",
      "name": "Llama 3",
      "vendor": "Meta AI",
      "release_date": "2024-04-18",
      "branch": "decoder_only",
      "parents": [
        "llama-2"
      ],
      "influences": [],
      "openness": "open_weights",
      "parameters": "8B, 70B",
      "context": "8K",
      "modalities": [
        "text"
      ],
      "capabilities": [
        "open_weights"
      ],
      "secondary_tags": [
        "open_weights",
        "historical_frontier"
      ],
      "notable": "Reset the open-weight quality bar; 8B beat Llama 2 70B on many benchmarks.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "llama-3-1",
      "name": "Llama 3.1",
      "vendor": "Meta AI",
      "release_date": "2024-07-23",
      "branch": "decoder_only",
      "parents": [
        "llama-3"
      ],
      "influences": [],
      "openness": "open_weights",
      "parameters": "8B, 70B, 405B",
      "context": "128K",
      "modalities": [
        "text"
      ],
      "capabilities": [
        "open_weights",
        "long_context"
      ],
      "secondary_tags": [
        "open_weights",
        "historical_frontier",
        "multilingual",
        "long_context"
      ],
      "notable": "405B closed the gap to GPT-4 in open weights.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "llama-3-2-edge",
      "name": "Llama 3.2 (1B / 3B edge)",
      "vendor": "Meta AI",
      "release_date": "2024-09-25",
      "branch": "decoder_only",
      "parents": [
        "llama-3-1"
      ],
      "influences": [],
      "openness": "open_weights",
      "parameters": "1B, 3B",
      "context": "128K",
      "modalities": [
        "text"
      ],
      "capabilities": [
        "open_weights",
        "long_context"
      ],
      "secondary_tags": [
        "open_weights",
        "small",
        "edge",
        "long_context"
      ],
      "notable": "First Meta entries in the on-device branch.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "llama-3-3-70b",
      "name": "Llama 3.3 70B",
      "vendor": "Meta AI",
      "release_date": "2024-12-06",
      "branch": "decoder_only",
      "parents": [
        "llama-3-1"
      ],
      "influences": [],
      "openness": "open_weights",
      "parameters": "70B",
      "context": "128K",
      "modalities": [
        "text"
      ],
      "capabilities": [
        "open_weights",
        "long_context"
      ],
      "secondary_tags": [
        "open_weights",
        "multilingual",
        "long_context"
      ],
      "notable": "Showed post-training alone could push 70B to near-405B quality.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "mistral-7b",
      "name": "Mistral 7B",
      "vendor": "Mistral AI",
      "release_date": "2023-09-27",
      "branch": "decoder_only",
      "parents": [
        "llama-2"
      ],
      "influences": [],
      "openness": "open_weights",
      "parameters": "7.3B",
      "context": "8K (sliding window 4K)",
      "modalities": [
        "text"
      ],
      "capabilities": [
        "open_weights"
      ],
      "secondary_tags": [
        "open_weights",
        "small",
        "eu"
      ],
      "notable": "Defined the open small-model frontier; Apache 2.0 set the licensing tone for European OSS LLMs.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "mistral-large-1",
      "name": "Mistral Large",
      "vendor": "Mistral AI",
      "release_date": "2024-02-26",
      "branch": "decoder_only",
      "parents": [
        "mistral-7b"
      ],
      "influences": [],
      "openness": "closed",
      "parameters": "undisclosed (~120B est.)",
      "context": "32K",
      "modalities": [
        "text"
      ],
      "capabilities": [],
      "secondary_tags": [
        "closed",
        "eu"
      ],
      "notable": "Marked Mistral's split into closed flagships and open mid-weights.\n",
      "status": "active",
      "placement_confidence": "medium"
    },
    {
      "id": "codestral-22b",
      "name": "Codestral 22B",
      "vendor": "Mistral AI",
      "release_date": "2024-05-29",
      "branch": "decoder_only",
      "parents": [
        "mistral-7b"
      ],
      "influences": [
        "code-llama"
      ],
      "openness": "mixed",
      "parameters": "22.2B",
      "context": "32K",
      "modalities": [
        "text"
      ],
      "capabilities": [],
      "secondary_tags": [
        "code_specialist",
        "eu"
      ],
      "notable": "Mistral's entry on the open-coder branch.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "mistral-nemo-12b",
      "name": "Mistral NeMo 12B",
      "vendor": "Mistral AI / NVIDIA",
      "release_date": "2024-07-18",
      "branch": "decoder_only",
      "parents": [
        "mistral-7b"
      ],
      "influences": [],
      "openness": "open_weights",
      "parameters": "12B",
      "context": "128K",
      "modalities": [
        "text"
      ],
      "capabilities": [
        "open_weights",
        "long_context"
      ],
      "secondary_tags": [
        "open_weights",
        "multilingual",
        "long_context",
        "eu"
      ],
      "notable": "Replaced Mistral 7B as the default open mid-weight.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "mistral-large-2",
      "name": "Mistral Large 2",
      "vendor": "Mistral AI",
      "release_date": "2024-07-24",
      "branch": "decoder_only",
      "parents": [
        "mistral-large-1"
      ],
      "influences": [],
      "openness": "open_weights",
      "parameters": "123B",
      "context": "128K",
      "modalities": [
        "text"
      ],
      "capabilities": [
        "open_weights",
        "long_context"
      ],
      "secondary_tags": [
        "open_weights",
        "historical_frontier",
        "long_context",
        "eu"
      ],
      "notable": "Mistral's strongest dense model; open-weight foundation for Pixtral Large.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "mistral-small-3",
      "name": "Mistral Small 3",
      "vendor": "Mistral AI",
      "release_date": "2025-01-30",
      "branch": "decoder_only",
      "parents": [
        "mistral-nemo-12b"
      ],
      "influences": [],
      "openness": "open_weights",
      "parameters": "24B",
      "context": "32K",
      "modalities": [
        "text"
      ],
      "capabilities": [
        "open_weights"
      ],
      "secondary_tags": [
        "open_weights",
        "small",
        "eu"
      ],
      "notable": "Mid-2020s reset of the Mistral small/medium tier.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "cohere-command",
      "name": "Cohere Command",
      "vendor": "Cohere",
      "release_date": "2023-09-29",
      "branch": "decoder_only",
      "parents": [],
      "influences": [],
      "openness": "closed",
      "parameters": "undisclosed (~52B est.)",
      "context": "4K",
      "modalities": [
        "text"
      ],
      "capabilities": [],
      "secondary_tags": [
        "closed",
        "enterprise"
      ],
      "notable": "Anchor for the Cohere open-weights branch.\n",
      "status": "active",
      "placement_confidence": "medium"
    },
    {
      "id": "command-r",
      "name": "Cohere Command R",
      "vendor": "Cohere",
      "release_date": "2024-03-11",
      "branch": "decoder_only",
      "parents": [
        "cohere-command"
      ],
      "influences": [],
      "openness": "open_weights",
      "parameters": "35B",
      "context": "128K",
      "modalities": [
        "text"
      ],
      "capabilities": [
        "open_weights",
        "long_context"
      ],
      "secondary_tags": [
        "open_weights",
        "rag_native",
        "multilingual",
        "enterprise",
        "long_context"
      ],
      "notable": "First open-weights model purpose-designed for RAG with grounded citations.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "command-r-plus",
      "name": "Cohere Command R+",
      "vendor": "Cohere",
      "release_date": "2024-04-04",
      "branch": "decoder_only",
      "parents": [
        "command-r"
      ],
      "influences": [],
      "openness": "open_weights",
      "parameters": "104B",
      "context": "128K",
      "modalities": [
        "text"
      ],
      "capabilities": [
        "tools",
        "open_weights",
        "long_context"
      ],
      "secondary_tags": [
        "open_weights",
        "rag_native",
        "multilingual",
        "enterprise",
        "historical_frontier",
        "long_context"
      ],
      "notable": "Largest open-weights enterprise/RAG model of early 2024.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "aya-23",
      "name": "Cohere Aya 23",
      "vendor": "Cohere For AI",
      "release_date": "2024-05-23",
      "branch": "decoder_only",
      "parents": [
        "command-r"
      ],
      "influences": [],
      "openness": "open_weights",
      "parameters": "8B, 35B",
      "context": "8K",
      "modalities": [
        "text"
      ],
      "capabilities": [
        "open_weights"
      ],
      "secondary_tags": [
        "open_weights",
        "multilingual"
      ],
      "notable": "Defined the multilingual open-weights frontier outside Llama.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "aya-expanse",
      "name": "Cohere Aya Expanse",
      "vendor": "Cohere For AI",
      "release_date": "2024-10-23",
      "branch": "decoder_only",
      "parents": [
        "aya-23"
      ],
      "influences": [],
      "openness": "open_weights",
      "parameters": "8B, 32B",
      "context": "128K",
      "modalities": [
        "text"
      ],
      "capabilities": [
        "open_weights",
        "long_context"
      ],
      "secondary_tags": [
        "open_weights",
        "multilingual",
        "long_context"
      ],
      "notable": "Long-context refresh of Cohere's multilingual line.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "command-a",
      "name": "Cohere Command A",
      "vendor": "Cohere",
      "release_date": "2025-03-11",
      "branch": "decoder_only",
      "parents": [
        "command-r-plus"
      ],
      "influences": [],
      "openness": "open_weights",
      "parameters": "111B",
      "context": "256K",
      "modalities": [
        "text"
      ],
      "capabilities": [
        "tools",
        "agentic",
        "open_weights",
        "long_context"
      ],
      "secondary_tags": [
        "open_weights",
        "rag_native",
        "agentic",
        "multilingual",
        "enterprise",
        "long_context"
      ],
      "notable": "Cohere's 2025 frontier; on-par with GPT-4o on agentic enterprise tasks.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "phi-1",
      "name": "Phi-1",
      "vendor": "Microsoft",
      "release_date": "2023-06-21",
      "branch": "decoder_only",
      "parents": [],
      "influences": [],
      "openness": "open_weights",
      "parameters": "1.3B",
      "context": "2048",
      "modalities": [
        "text"
      ],
      "capabilities": [
        "open_weights"
      ],
      "secondary_tags": [
        "open_weights",
        "small",
        "distilled",
        "code_specialist"
      ],
      "notable": "Showed quality-greater-than-scale for narrow domains; SOTA on HumanEval/MBPP for sub-2B models.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "phi-1-5",
      "name": "Phi-1.5",
      "vendor": "Microsoft",
      "release_date": "2023-09-11",
      "branch": "decoder_only",
      "parents": [
        "phi-1"
      ],
      "influences": [],
      "openness": "open_weights",
      "parameters": "1.3B",
      "context": "2048",
      "modalities": [
        "text"
      ],
      "capabilities": [
        "open_weights"
      ],
      "secondary_tags": [
        "open_weights",
        "small",
        "distilled"
      ],
      "notable": "Matched models 5x its size on common-sense reasoning.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "phi-2",
      "name": "Phi-2",
      "vendor": "Microsoft",
      "release_date": "2023-12-12",
      "branch": "decoder_only",
      "parents": [
        "phi-1-5"
      ],
      "influences": [],
      "openness": "open_weights",
      "parameters": "2.7B",
      "context": "2048",
      "modalities": [
        "text"
      ],
      "capabilities": [
        "open_weights"
      ],
      "secondary_tags": [
        "open_weights",
        "small",
        "distilled"
      ],
      "notable": "Matched or beat models up to 25x larger on complex benchmarks; flagship of the SML thesis.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "phi-3",
      "name": "Phi-3 (mini / small / medium)",
      "vendor": "Microsoft",
      "release_date": "2024-04-23",
      "branch": "decoder_only",
      "parents": [
        "phi-2"
      ],
      "influences": [],
      "openness": "open_weights",
      "parameters": "3.8B (mini) / 7B (small) / 14B (medium)",
      "context": "4K / 128K",
      "modalities": [
        "text"
      ],
      "capabilities": [
        "open_weights",
        "long_context"
      ],
      "secondary_tags": [
        "open_weights",
        "small",
        "distilled",
        "edge",
        "long_context"
      ],
      "notable": "Phi-3-mini ran on a phone, comparable to GPT-3.5.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "phi-4",
      "name": "Phi-4",
      "vendor": "Microsoft",
      "release_date": "2024-12-12",
      "branch": "decoder_only",
      "parents": [
        "phi-3"
      ],
      "influences": [],
      "openness": "open_weights",
      "parameters": "14B",
      "context": "16K",
      "modalities": [
        "text"
      ],
      "capabilities": [
        "open_weights"
      ],
      "secondary_tags": [
        "open_weights",
        "distilled"
      ],
      "notable": "Outperformed GPT-4o on competition math; MIT-licensed weights.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "granite-3-0",
      "name": "IBM Granite 3.0",
      "vendor": "IBM",
      "release_date": "2024-10-21",
      "branch": "decoder_only",
      "parents": [],
      "influences": [
        "llama-2"
      ],
      "openness": "open_weights",
      "parameters": "8B, 2B dense; 3B-A800M, 1B-A400M MoE",
      "context": "4K",
      "modalities": [
        "text"
      ],
      "capabilities": [
        "open_weights"
      ],
      "secondary_tags": [
        "open_weights",
        "enterprise",
        "multilingual",
        "code_specialist"
      ],
      "notable": "IBM's reset of the Granite line into a credible enterprise open-weights family.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "granite-3-1",
      "name": "IBM Granite 3.1",
      "vendor": "IBM",
      "release_date": "2025-01-12",
      "branch": "decoder_only",
      "parents": [
        "granite-3-0"
      ],
      "influences": [],
      "openness": "open_weights",
      "parameters": "2B, 8B dense; 1B-A400M, 3B-A800M MoE",
      "context": "128K",
      "modalities": [
        "text"
      ],
      "capabilities": [
        "open_weights",
        "long_context"
      ],
      "secondary_tags": [
        "open_weights",
        "enterprise",
        "long_context"
      ],
      "notable": "Long-context refresh of Granite.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "granite-3-3",
      "name": "IBM Granite 3.3",
      "vendor": "IBM",
      "release_date": "2025-04-16",
      "branch": "decoder_only",
      "parents": [
        "granite-3-2"
      ],
      "influences": [],
      "openness": "open_weights",
      "parameters": "2B, 8B",
      "context": "128K",
      "modalities": [
        "text"
      ],
      "capabilities": [
        "open_weights",
        "long_context"
      ],
      "secondary_tags": [
        "open_weights",
        "enterprise",
        "multilingual",
        "code_specialist",
        "long_context"
      ],
      "notable": "Last dense-Transformer Granite before the 4.0 hybrid pivot.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "falcon-7b-40b",
      "name": "Falcon 7B / 40B",
      "vendor": "TII",
      "release_date": "2023-05-25",
      "branch": "decoder_only",
      "parents": [],
      "influences": [],
      "openness": "open_weights",
      "parameters": "7B, 40B",
      "context": "2K",
      "modalities": [
        "text"
      ],
      "capabilities": [
        "open_weights"
      ],
      "secondary_tags": [
        "open_weights",
        "uae"
      ],
      "notable": "First Apache-2.0 non-Western frontier open-weight.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "falcon-180b",
      "name": "Falcon 180B",
      "vendor": "TII",
      "release_date": "2023-09-06",
      "branch": "decoder_only",
      "parents": [
        "falcon-7b-40b"
      ],
      "influences": [],
      "openness": "open_weights",
      "parameters": "180B",
      "context": "2K",
      "modalities": [
        "text"
      ],
      "capabilities": [
        "open_weights"
      ],
      "secondary_tags": [
        "open_weights",
        "historical_frontier",
        "uae"
      ],
      "notable": "Largest publicly-available open-weight model of 2023 before Llama 3.1 405B.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "falcon-3",
      "name": "Falcon 3 family",
      "vendor": "TII",
      "release_date": "2024-12-17",
      "branch": "decoder_only",
      "parents": [
        "falcon-180b"
      ],
      "influences": [],
      "openness": "open_weights",
      "parameters": "1B, 3B, 7B, 10B",
      "context": "32K",
      "modalities": [
        "text"
      ],
      "capabilities": [
        "open_weights"
      ],
      "secondary_tags": [
        "open_weights",
        "small",
        "code_specialist",
        "uae"
      ],
      "notable": "Pivoted Falcon to Llama-compatible small/efficient lineup.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "qwen-7b",
      "name": "Qwen-7B",
      "vendor": "Alibaba",
      "release_date": "2023-08",
      "branch": "decoder_only",
      "parents": [],
      "influences": [
        "llama-2"
      ],
      "openness": "open_weights",
      "parameters": "7B",
      "context": "32K",
      "modalities": [
        "text"
      ],
      "capabilities": [
        "open_weights"
      ],
      "secondary_tags": [
        "open_weights",
        "bilingual",
        "china"
      ],
      "notable": "Alibaba's debut on the open-weights frontier.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "qwen-72b",
      "name": "Qwen 1.8B / 14B / 72B",
      "vendor": "Alibaba",
      "release_date": "2023-12",
      "branch": "decoder_only",
      "parents": [
        "qwen-7b"
      ],
      "influences": [],
      "openness": "open_weights",
      "parameters": "1.8B / 14B / 72B",
      "context": "32K",
      "modalities": [
        "text"
      ],
      "capabilities": [
        "open_weights"
      ],
      "secondary_tags": [
        "open_weights",
        "china"
      ],
      "notable": "Qwen reaches the 72B frontier-adjacent open-weights tier.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "qwen-1-5",
      "name": "Qwen 1.5",
      "vendor": "Alibaba",
      "release_date": "2024-02",
      "branch": "decoder_only",
      "parents": [
        "qwen-72b"
      ],
      "influences": [],
      "openness": "open_weights",
      "parameters": "0.5B / 1.8B / 4B / 7B / 14B / 32B / 72B / 110B",
      "context": "32K",
      "modalities": [
        "text"
      ],
      "capabilities": [
        "open_weights"
      ],
      "secondary_tags": [
        "open_weights",
        "china"
      ],
      "notable": "Quietly shipped MoE in Feb 2024, predating DeepSeek-V2.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "qwen-2",
      "name": "Qwen2",
      "vendor": "Alibaba",
      "release_date": "2024-06",
      "branch": "decoder_only",
      "parents": [
        "qwen-1-5"
      ],
      "influences": [],
      "openness": "open_weights",
      "parameters": "0.5B-72B",
      "context": "128K (72B)",
      "modalities": [
        "text"
      ],
      "capabilities": [
        "open_weights",
        "long_context"
      ],
      "secondary_tags": [
        "open_weights",
        "long_context",
        "china"
      ],
      "notable": "Continued Alibaba's open-weights momentum.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "qwen-2-5",
      "name": "Qwen2.5",
      "vendor": "Alibaba",
      "release_date": "2024-09",
      "branch": "decoder_only",
      "parents": [
        "qwen-2"
      ],
      "influences": [],
      "openness": "open_weights",
      "parameters": "0.5B / 1.5B / 3B / 7B / 14B / 32B / 72B",
      "context": "128K",
      "modalities": [
        "text"
      ],
      "capabilities": [
        "open_weights",
        "long_context"
      ],
      "secondary_tags": [
        "open_weights",
        "code_specialist",
        "math_specialist",
        "long_context",
        "china"
      ],
      "notable": "Strong open-weights baseline for the Chinese frontier.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "deepseek-llm-67b",
      "name": "DeepSeek LLM 67B",
      "vendor": "DeepSeek AI",
      "release_date": "2023-11-29",
      "branch": "decoder_only",
      "parents": [],
      "influences": [
        "llama-2"
      ],
      "openness": "open_weights",
      "parameters": "67B",
      "context": "4K",
      "modalities": [
        "text"
      ],
      "capabilities": [
        "open_weights"
      ],
      "secondary_tags": [
        "open_weights",
        "china"
      ],
      "notable": "DeepSeek's debut; outperformed Llama-2-70B on code and math.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "deepseek-coder",
      "name": "DeepSeek-Coder",
      "vendor": "DeepSeek AI",
      "release_date": "2023-11",
      "branch": "decoder_only",
      "parents": [
        "deepseek-llm-67b"
      ],
      "influences": [],
      "openness": "open_weights",
      "parameters": "1.3B / 6.7B / 33B",
      "context": "16K",
      "modalities": [
        "text"
      ],
      "capabilities": [
        "open_weights"
      ],
      "secondary_tags": [
        "open_weights",
        "code_specialist",
        "china"
      ],
      "notable": "First DeepSeek code-specialized model.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "deepseek-math",
      "name": "DeepSeekMath 7B",
      "vendor": "DeepSeek AI",
      "release_date": "2024-02",
      "branch": "decoder_only",
      "parents": [
        "deepseek-coder"
      ],
      "influences": [],
      "openness": "open_weights",
      "parameters": "7B",
      "context": "4K",
      "modalities": [
        "text"
      ],
      "capabilities": [
        "open_weights"
      ],
      "secondary_tags": [
        "open_weights",
        "math_specialist",
        "china"
      ],
      "notable": "Pioneered Group Relative Policy Optimization (GRPO), later used in R1.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "yi-34b",
      "name": "Yi-34B",
      "vendor": "01.AI",
      "release_date": "2023-11",
      "branch": "decoder_only",
      "parents": [],
      "influences": [
        "llama-2"
      ],
      "openness": "open_weights",
      "parameters": "6B, 34B",
      "context": "4K (extended to 200K variants)",
      "modalities": [
        "text"
      ],
      "capabilities": [
        "open_weights"
      ],
      "secondary_tags": [
        "open_weights",
        "bilingual",
        "china"
      ],
      "notable": "01.AI's open-weights debut.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "yi-1-5",
      "name": "Yi-1.5",
      "vendor": "01.AI",
      "release_date": "2024-05-13",
      "branch": "decoder_only",
      "parents": [
        "yi-34b"
      ],
      "influences": [],
      "openness": "open_weights",
      "parameters": "6B / 9B / 34B",
      "context": "4K (extendable)",
      "modalities": [
        "text"
      ],
      "capabilities": [
        "open_weights"
      ],
      "secondary_tags": [
        "open_weights",
        "china"
      ],
      "notable": "Refined the Yi-34B baseline.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "ernie-4",
      "name": "ERNIE 4.0",
      "vendor": "Baidu",
      "release_date": "2023-10",
      "branch": "decoder_only",
      "parents": [],
      "influences": [],
      "openness": "closed",
      "parameters": "undisclosed",
      "context": "8K",
      "modalities": [
        "text"
      ],
      "capabilities": [],
      "secondary_tags": [
        "closed",
        "multilingual",
        "china"
      ],
      "notable": "China's GPT-4 peer at the time.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "grok-1",
      "name": "Grok-1",
      "vendor": "xAI",
      "release_date": "2023-11-04",
      "branch": "decoder_only",
      "parents": [],
      "influences": [],
      "openness": "open_weights",
      "parameters": "314B (MoE, 25% active)",
      "context": "8K",
      "modalities": [
        "text"
      ],
      "capabilities": [
        "open_weights"
      ],
      "secondary_tags": [
        "open_weights",
        "mixture_of_experts"
      ],
      "notable": "xAI's open-weights debut.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "grok-1-5",
      "name": "Grok-1.5",
      "vendor": "xAI",
      "release_date": "2024-03",
      "branch": "decoder_only",
      "parents": [
        "grok-1"
      ],
      "influences": [],
      "openness": "closed",
      "parameters": "undisclosed",
      "context": "128K",
      "modalities": [
        "text"
      ],
      "capabilities": [
        "long_context"
      ],
      "secondary_tags": [
        "closed",
        "long_context"
      ],
      "notable": "Bridge to the multimodal Grok-1.5V.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "grok-2",
      "name": "Grok-2",
      "vendor": "xAI",
      "release_date": "2024-08",
      "branch": "decoder_only",
      "parents": [
        "grok-1-5"
      ],
      "influences": [],
      "openness": "closed",
      "parameters": "undisclosed",
      "context": "128K",
      "modalities": [
        "text",
        "image"
      ],
      "capabilities": [
        "tools",
        "multimodal"
      ],
      "secondary_tags": [
        "closed",
        "multimodal",
        "tool_using"
      ],
      "notable": "Last pre-reasoning xAI flagship.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "jurassic-2",
      "name": "AI21 Jurassic-2",
      "vendor": "AI21 Labs",
      "release_date": "2023-03-09",
      "branch": "decoder_only",
      "parents": [],
      "influences": [],
      "openness": "closed",
      "parameters": "Light 7B, Mid 17B, Ultra 60B",
      "context": "8K",
      "modalities": [
        "text"
      ],
      "capabilities": [],
      "secondary_tags": [
        "closed",
        "multilingual"
      ],
      "notable": "Pre-Jamba dense lineage.\n",
      "status": "active",
      "placement_confidence": "medium"
    },
    {
      "id": "jamba",
      "name": "AI21 Jamba v0.1",
      "vendor": "AI21 Labs",
      "release_date": "2024-03-28",
      "branch": "decoder_only",
      "parents": [
        "jurassic-2"
      ],
      "influences": [],
      "openness": "open_weights",
      "parameters": "52B total / 12B active",
      "context": "256K",
      "modalities": [
        "text"
      ],
      "capabilities": [
        "open_weights",
        "long_context"
      ],
      "secondary_tags": [
        "open_weights",
        "ssm_hybrid",
        "mamba",
        "mixture_of_experts",
        "long_context"
      ],
      "notable": "First production-grade open SSM-Transformer hybrid.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "jamba-1-5",
      "name": "AI21 Jamba 1.5",
      "vendor": "AI21 Labs",
      "release_date": "2024-08-22",
      "branch": "decoder_only",
      "parents": [
        "jamba"
      ],
      "influences": [],
      "openness": "open_weights",
      "parameters": "Mini 52B/12B active; Large 398B/94B active",
      "context": "256K",
      "modalities": [
        "text"
      ],
      "capabilities": [
        "open_weights",
        "long_context"
      ],
      "secondary_tags": [
        "open_weights",
        "ssm_hybrid",
        "mamba",
        "mixture_of_experts",
        "long_context"
      ],
      "notable": "Scaled Jamba SSM-hybrid into a frontier-class open model.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "jamba-2",
      "name": "AI21 Jamba 2",
      "vendor": "AI21 Labs",
      "release_date": "2026-01-08",
      "branch": "decoder_only",
      "parents": [
        "jamba-1-5"
      ],
      "influences": [],
      "openness": "open_weights",
      "parameters": "Large 398B/94B; Mini 52B/12B; 3B",
      "context": "256K",
      "modalities": [
        "text"
      ],
      "capabilities": [
        "open_weights",
        "long_context"
      ],
      "secondary_tags": [
        "open_weights",
        "ssm_hybrid",
        "mamba",
        "mixture_of_experts",
        "long_context"
      ],
      "notable": "Latest evolution of the Western SSM-hybrid branch.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "glm-4",
      "name": "GLM-4",
      "vendor": "Z.AI (Zhipu)",
      "release_date": "2024-01",
      "branch": "decoder_only",
      "parents": [],
      "influences": [],
      "openness": "closed",
      "parameters": "Undisclosed flagship",
      "context": "128K",
      "modalities": [
        "text"
      ],
      "capabilities": [
        "long_context"
      ],
      "secondary_tags": [
        "closed",
        "china",
        "long_context"
      ],
      "notable": "Anchor for the GLM-Z1 reasoning fork and GLM-5 MoE family.\n",
      "status": "active",
      "placement_confidence": "medium"
    },
    {
      "id": "glm-4-32b-0414",
      "name": "GLM-4-32B-0414",
      "vendor": "Z.AI (Zhipu)",
      "release_date": "2025-04",
      "branch": "decoder_only",
      "parents": [
        "glm-4"
      ],
      "influences": [],
      "openness": "open_weights",
      "parameters": "32B",
      "context": "128K",
      "modalities": [
        "text"
      ],
      "capabilities": [
        "open_weights",
        "long_context"
      ],
      "secondary_tags": [
        "open_weights",
        "china",
        "long_context"
      ],
      "notable": "Zhipu's open-weights baseline for the GLM-Z1 reasoning fork.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "kimi-k1",
      "name": "Kimi K1",
      "vendor": "Moonshot AI",
      "release_date": "2024-03",
      "branch": "decoder_only",
      "parents": [],
      "influences": [],
      "openness": "closed",
      "parameters": "Undisclosed",
      "context": "200K (chat product)",
      "modalities": [
        "text"
      ],
      "capabilities": [
        "long_context"
      ],
      "secondary_tags": [
        "closed",
        "china",
        "long_context"
      ],
      "notable": "Origin of the Kimi line.\n",
      "status": "active",
      "placement_confidence": "medium"
    },
    {
      "id": "switch-transformer",
      "name": "Switch Transformer",
      "vendor": "Google",
      "release_date": "2021-01-11",
      "branch": "mixture_of_experts",
      "parents": [
        "t5"
      ],
      "influences": [],
      "openness": "open_source",
      "parameters": "Up to 1.6T total (sparsely activated)",
      "context": "512",
      "modalities": [
        "text"
      ],
      "capabilities": [],
      "secondary_tags": [
        "open_source",
        "foundational"
      ],
      "notable": "First trillion-parameter LM; defined the modern MoE pattern at scale.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "glam",
      "name": "GLaM",
      "vendor": "Google",
      "release_date": "2021-12-09",
      "branch": "mixture_of_experts",
      "parents": [],
      "influences": [
        "switch-transformer"
      ],
      "openness": "closed",
      "parameters": "1.2T total / 97B active per token",
      "context": "1024",
      "modalities": [
        "text"
      ],
      "capabilities": [],
      "secondary_tags": [
        "closed"
      ],
      "notable": "Demonstrated MoE works for decoder-only generation.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "st-moe",
      "name": "ST-MoE",
      "vendor": "Google",
      "release_date": "2022-02-17",
      "branch": "mixture_of_experts",
      "parents": [
        "switch-transformer"
      ],
      "influences": [],
      "openness": "closed",
      "parameters": "269B total (32B compute-equivalent)",
      "context": "1024",
      "modalities": [
        "text"
      ],
      "capabilities": [],
      "secondary_tags": [
        "closed"
      ],
      "notable": "First sparse model SOTA on transfer learning; informed downstream MoE in Gemini.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "mixtral-8x7b",
      "name": "Mixtral 8x7B",
      "vendor": "Mistral AI",
      "release_date": "2023-12-11",
      "branch": "mixture_of_experts",
      "parents": [
        "mistral-7b"
      ],
      "influences": [
        "glam"
      ],
      "openness": "open_weights",
      "parameters": "46.7B total / 12.9B active",
      "context": "32K",
      "modalities": [
        "text"
      ],
      "capabilities": [
        "open_weights"
      ],
      "secondary_tags": [
        "open_weights",
        "eu",
        "long_context"
      ],
      "notable": "First widely-deployed open-weights MoE; defined the open-MoE branch.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "mixtral-8x22b",
      "name": "Mixtral 8x22B",
      "vendor": "Mistral AI",
      "release_date": "2024-04-10",
      "branch": "mixture_of_experts",
      "parents": [
        "mixtral-8x7b"
      ],
      "influences": [],
      "openness": "open_weights",
      "parameters": "141B total / 39B active",
      "context": "64K",
      "modalities": [
        "text"
      ],
      "capabilities": [
        "open_weights",
        "long_context"
      ],
      "secondary_tags": [
        "open_weights",
        "multilingual",
        "eu",
        "long_context"
      ],
      "notable": "Largest fully Apache-2.0 MoE before DBRX/Arctic.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "dbrx",
      "name": "DBRX",
      "vendor": "Databricks",
      "release_date": "2024-03-27",
      "branch": "mixture_of_experts",
      "parents": [
        "mixtral-8x7b"
      ],
      "influences": [],
      "openness": "open_weights",
      "parameters": "132B total / 36B active",
      "context": "32K",
      "modalities": [
        "text"
      ],
      "capabilities": [
        "open_weights"
      ],
      "secondary_tags": [
        "open_weights",
        "enterprise"
      ],
      "notable": "Pushed fine-grained MoE into the open; influenced Llama 4 routing design.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "snowflake-arctic",
      "name": "Snowflake Arctic",
      "vendor": "Snowflake",
      "release_date": "2024-04-24",
      "branch": "mixture_of_experts",
      "parents": [
        "dbrx",
        "mixtral-8x22b"
      ],
      "influences": [],
      "openness": "open_weights",
      "parameters": "480B total / 17B active",
      "context": "4K (initial)",
      "modalities": [
        "text"
      ],
      "capabilities": [
        "open_weights"
      ],
      "secondary_tags": [
        "open_weights",
        "enterprise"
      ],
      "notable": "Largest open-weights MoE of 2024 by total params.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "phi-3-5-moe",
      "name": "Phi-3.5 MoE",
      "vendor": "Microsoft",
      "release_date": "2024-08-21",
      "branch": "mixture_of_experts",
      "parents": [
        "phi-3"
      ],
      "influences": [
        "mixtral-8x7b"
      ],
      "openness": "open_weights",
      "parameters": "41.9B total / 6.6B active (16 experts)",
      "context": "128K",
      "modalities": [
        "text"
      ],
      "capabilities": [
        "open_weights",
        "long_context"
      ],
      "secondary_tags": [
        "open_weights",
        "small",
        "distilled",
        "long_context"
      ],
      "notable": "Brought MoE into the small-model branch.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "deepseek-v2",
      "name": "DeepSeek-V2",
      "vendor": "DeepSeek AI",
      "release_date": "2024-05-06",
      "branch": "mixture_of_experts",
      "parents": [
        "deepseek-llm-67b"
      ],
      "influences": [],
      "openness": "open_weights",
      "parameters": "236B total / 21B active",
      "context": "128K",
      "modalities": [
        "text"
      ],
      "capabilities": [
        "open_weights",
        "long_context"
      ],
      "secondary_tags": [
        "open_weights",
        "china",
        "long_context"
      ],
      "notable": "Inflection point — MLA plus DeepSeekMoE rewired Chinese training economics.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "deepseek-coder-v2",
      "name": "DeepSeek-Coder-V2",
      "vendor": "DeepSeek AI",
      "release_date": "2024-06",
      "branch": "mixture_of_experts",
      "parents": [
        "deepseek-v2"
      ],
      "influences": [],
      "openness": "open_weights",
      "parameters": "236B/21B (Std); 16B/2.4B (Lite)",
      "context": "128K",
      "modalities": [
        "text"
      ],
      "capabilities": [
        "open_weights",
        "long_context"
      ],
      "secondary_tags": [
        "open_weights",
        "code_specialist",
        "china",
        "long_context"
      ],
      "notable": "Extended DeepSeek-V2 to coding workloads.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "deepseek-v3",
      "name": "DeepSeek-V3",
      "vendor": "DeepSeek AI",
      "release_date": "2024-12-26",
      "branch": "mixture_of_experts",
      "parents": [
        "deepseek-v2"
      ],
      "influences": [],
      "openness": "open_weights",
      "parameters": "671B total / 37B active",
      "context": "128K",
      "modalities": [
        "text"
      ],
      "capabilities": [
        "open_weights",
        "long_context"
      ],
      "secondary_tags": [
        "open_weights",
        "historical_frontier",
        "china",
        "long_context"
      ],
      "notable": "The cost-shock release that reframed economics of frontier training.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "deepseek-v3-0324",
      "name": "DeepSeek-V3-0324",
      "vendor": "DeepSeek AI",
      "release_date": "2025-03-24",
      "branch": "mixture_of_experts",
      "parents": [
        "deepseek-v3"
      ],
      "influences": [
        "deepseek-r1"
      ],
      "openness": "open_weights",
      "parameters": "671B total / 37B active",
      "context": "128K",
      "modalities": [
        "text"
      ],
      "capabilities": [
        "open_weights",
        "long_context"
      ],
      "secondary_tags": [
        "open_weights",
        "china",
        "long_context"
      ],
      "notable": "First DeepSeek refresh after R1 propagated reasoning traits back into the base.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "deepseek-v3-2",
      "name": "DeepSeek-V3.2",
      "vendor": "DeepSeek AI",
      "release_date": "2025-12",
      "branch": "mixture_of_experts",
      "parents": [
        "deepseek-v3-0324"
      ],
      "influences": [],
      "openness": "open_weights",
      "parameters": "671B total / 37B active",
      "context": "128K",
      "modalities": [
        "text"
      ],
      "capabilities": [
        "open_weights",
        "long_context"
      ],
      "secondary_tags": [
        "open_weights",
        "china",
        "long_context"
      ],
      "notable": "Introduced DSA, the first cross-vendor architectural pattern to propagate within the Chinese frontier.\n",
      "status": "active",
      "placement_confidence": "medium"
    },
    {
      "id": "deepseek-v4-pro",
      "name": "DeepSeek-V4 Pro",
      "vendor": "DeepSeek AI",
      "release_date": "2026-04-24",
      "branch": "mixture_of_experts",
      "parents": [
        "deepseek-v3-2"
      ],
      "influences": [],
      "openness": "open_weights",
      "parameters": "1.6T total / 49B active",
      "context": "1M",
      "modalities": [
        "text"
      ],
      "capabilities": [
        "tools",
        "agentic",
        "open_weights",
        "long_context"
      ],
      "secondary_tags": [
        "open_weights",
        "frontier",
        "china",
        "ultra_long_context",
        "agentic"
      ],
      "notable": "First Chinese open-weights model to match Western frontier on aggregate benchmarks (BenchLM 87). Updated 2026-08-12 for the 0813 GA build, which is where the openness label starts to strain: DeepSeek reports large gains over the April preview (Terminal Bench 2.1 72.1 to 87.9, DeepSWE 12.8 to 62.7, CyberGym 52.7 to 83.3, AA Intelligence Index 45 to 53), but the Hugging Face repositories still hosted the April preview weights after the swap. For a period, every independent evaluation of \"DeepSeek V4 Pro\" was measuring a build no one could download, and every downloadable build was one no one was serving. A price change was also announced for 2026-08-16 replacing the flat $0.435 input / $0.87 output rate with peak and off-peak tiers reported at $1.32 / $3.96 peak and half that off-peak.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "deepseek-v4-flash",
      "name": "DeepSeek-V4 Flash",
      "vendor": "DeepSeek AI",
      "release_date": "2026-04-24",
      "branch": "mixture_of_experts",
      "parents": [
        "deepseek-v3-2"
      ],
      "influences": [],
      "openness": "open_weights",
      "parameters": "284B total / 13B active",
      "context": "1M",
      "modalities": [
        "text"
      ],
      "capabilities": [
        "tools",
        "agentic",
        "open_weights",
        "long_context"
      ],
      "secondary_tags": [
        "open_weights",
        "china",
        "ultra_long_context",
        "agentic"
      ],
      "notable": "Cost-efficient companion to V4 Pro at consumer-deployable scale.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "qwen-3",
      "name": "Qwen3 (235B-A22B / 30B-A3B)",
      "vendor": "Alibaba",
      "release_date": "2025-04",
      "branch": "mixture_of_experts",
      "parents": [
        "qwen-2-5"
      ],
      "influences": [
        "deepseek-v3",
        "deepseek-r1"
      ],
      "openness": "open_weights",
      "parameters": "235B/22B (flagship MoE); 30B/3B; dense 0.6B-32B",
      "context": "128K",
      "modalities": [
        "text"
      ],
      "capabilities": [
        "open_weights",
        "long_context"
      ],
      "secondary_tags": [
        "open_weights",
        "hybrid_reasoning",
        "frontier",
        "china",
        "long_context"
      ],
      "notable": "First major Western/Eastern model with a hybrid thinking-mode toggle.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "kimi-k2",
      "name": "Kimi K2",
      "vendor": "Moonshot AI",
      "release_date": "2025-07",
      "branch": "mixture_of_experts",
      "parents": [
        "kimi-k1-5"
      ],
      "influences": [
        "deepseek-v3"
      ],
      "openness": "open_weights",
      "parameters": "1T total / 32B active",
      "context": "128K",
      "modalities": [
        "text"
      ],
      "capabilities": [
        "tools",
        "agentic",
        "open_weights",
        "long_context"
      ],
      "secondary_tags": [
        "open_weights",
        "agentic",
        "china",
        "long_context"
      ],
      "notable": "First open-weights trillion-parameter MoE.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "kimi-k2-6",
      "name": "Kimi K2.6",
      "vendor": "Moonshot AI",
      "release_date": "2026-04-20",
      "branch": "mixture_of_experts",
      "parents": [
        "kimi-k2"
      ],
      "influences": [],
      "openness": "open_weights",
      "parameters": "1T total / 32B active (384 routed experts: 8 active + 1 shared)",
      "context": "262K",
      "modalities": [
        "text",
        "image"
      ],
      "capabilities": [
        "tools",
        "agentic",
        "multi_agent_native",
        "multimodal",
        "open_weights",
        "long_context"
      ],
      "secondary_tags": [
        "open_weights",
        "agentic",
        "multi_agent_native",
        "frontier",
        "china",
        "multimodal",
        "long_context"
      ],
      "notable": "Native 300-sub-agent swarm orchestration; 58.6% SWE-Bench Pro.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "glm-5",
      "name": "GLM-5",
      "vendor": "Z.AI (Zhipu)",
      "release_date": "2026-02",
      "branch": "mixture_of_experts",
      "parents": [
        "glm-4-32b-0414"
      ],
      "influences": [
        "deepseek-v3"
      ],
      "openness": "open_weights",
      "parameters": "744B total / 40B active",
      "context": "200K",
      "modalities": [
        "text"
      ],
      "capabilities": [
        "open_weights",
        "long_context"
      ],
      "secondary_tags": [
        "open_weights",
        "frontier",
        "china",
        "huawei_trained",
        "long_context"
      ],
      "notable": "First frontier-tier Chinese model trained without NVIDIA hardware.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "hunyuan-large",
      "name": "Hunyuan-Large",
      "vendor": "Tencent",
      "release_date": "2024-11",
      "branch": "mixture_of_experts",
      "parents": [],
      "influences": [],
      "openness": "open_weights",
      "parameters": "389B total / 52B active",
      "context": "256K",
      "modalities": [
        "text"
      ],
      "capabilities": [
        "open_weights",
        "long_context"
      ],
      "secondary_tags": [
        "open_weights",
        "china",
        "long_context"
      ],
      "notable": "Tencent's open-weights frontier debut.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "hunyuan-turbos",
      "name": "Hunyuan-TurboS",
      "vendor": "Tencent",
      "release_date": "2025-03",
      "branch": "mixture_of_experts",
      "parents": [
        "hunyuan-large"
      ],
      "influences": [],
      "openness": "closed",
      "parameters": "560B total / 56B active",
      "context": "256K",
      "modalities": [
        "text"
      ],
      "capabilities": [
        "long_context"
      ],
      "secondary_tags": [
        "closed",
        "ssm_hybrid",
        "mamba",
        "china",
        "long_context"
      ],
      "notable": "First ultra-large SSM-hybrid MoE.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "hunyuan-2",
      "name": "Hunyuan 2.0",
      "vendor": "Tencent",
      "release_date": "2025-12",
      "branch": "mixture_of_experts",
      "parents": [
        "hunyuan-turbos"
      ],
      "influences": [],
      "openness": "closed",
      "parameters": "406B total / 32B active",
      "context": "256K",
      "modalities": [
        "text"
      ],
      "capabilities": [
        "long_context"
      ],
      "secondary_tags": [
        "closed",
        "china",
        "long_context"
      ],
      "notable": "Continued the Hunyuan MoE line.\n",
      "status": "active",
      "placement_confidence": "medium"
    },
    {
      "id": "minimax-text-01",
      "name": "MiniMax-Text-01",
      "vendor": "MiniMax",
      "release_date": "2025-01-15",
      "branch": "mixture_of_experts",
      "parents": [],
      "influences": [],
      "openness": "open_weights",
      "parameters": "456B total / 45.9B active",
      "context": "1M (4M extrapolation)",
      "modalities": [
        "text"
      ],
      "capabilities": [
        "open_weights",
        "long_context"
      ],
      "secondary_tags": [
        "open_weights",
        "china",
        "ultra_long_context"
      ],
      "notable": "Pioneered Lightning Attention at frontier scale.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "gpt-oss-120b",
      "name": "gpt-oss-120b",
      "vendor": "OpenAI",
      "release_date": "2025-08-05",
      "branch": "mixture_of_experts",
      "parents": [],
      "influences": [
        "gpt-3",
        "o4-mini"
      ],
      "openness": "open_weights",
      "parameters": "116.8B total / 5.1B active per token",
      "context": "128K",
      "modalities": [
        "text"
      ],
      "capabilities": [
        "tools",
        "agentic",
        "open_weights",
        "long_context"
      ],
      "secondary_tags": [
        "open_weights",
        "agentic",
        "long_context"
      ],
      "notable": "OpenAI's first open-weight release since GPT-2 (2019). Apache 2.0. Near-parity with o4-mini on a single 80GB GPU.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "gpt-oss-20b",
      "name": "gpt-oss-20b",
      "vendor": "OpenAI",
      "release_date": "2025-08-05",
      "branch": "mixture_of_experts",
      "parents": [
        "gpt-oss-120b"
      ],
      "influences": [
        "o3-mini"
      ],
      "openness": "open_weights",
      "parameters": "20.9B total / 3.6B active per token",
      "context": "128K",
      "modalities": [
        "text"
      ],
      "capabilities": [
        "tools",
        "agentic",
        "open_weights",
        "long_context"
      ],
      "secondary_tags": [
        "open_weights",
        "small",
        "edge",
        "agentic",
        "long_context"
      ],
      "notable": "First time OpenAI shipped a frontier-style reasoning model that fits on a laptop GPU.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "llama-4-scout",
      "name": "Llama 4 Scout",
      "vendor": "Meta AI",
      "release_date": "2025-04-05",
      "branch": "mixture_of_experts",
      "parents": [
        "llama-3-3-70b"
      ],
      "influences": [
        "mixtral-8x7b"
      ],
      "openness": "open_weights",
      "parameters": "109B total / 17B active",
      "context": "10M (claimed)",
      "modalities": [
        "text",
        "image",
        "video"
      ],
      "capabilities": [
        "tools",
        "multimodal",
        "open_weights",
        "long_context"
      ],
      "secondary_tags": [
        "open_weights",
        "multimodal",
        "frontier",
        "ultra_long_context"
      ],
      "notable": "First open-weights model with a 10M-token context window; Meta's pivot to MoE.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "llama-4-maverick",
      "name": "Llama 4 Maverick",
      "vendor": "Meta AI",
      "release_date": "2025-04-05",
      "branch": "mixture_of_experts",
      "parents": [
        "llama-3-3-70b"
      ],
      "influences": [
        "mixtral-8x22b",
        "dbrx"
      ],
      "openness": "open_weights",
      "parameters": "400B total / 17B active",
      "context": "1M",
      "modalities": [
        "text",
        "image",
        "video"
      ],
      "capabilities": [
        "tools",
        "multimodal",
        "open_weights",
        "long_context"
      ],
      "secondary_tags": [
        "open_weights",
        "multimodal",
        "frontier",
        "long_context"
      ],
      "notable": "Largest open-weights MoE released by Meta; replaced dense flagship strategy.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "mistral-large-3",
      "name": "Mistral Large 3",
      "vendor": "Mistral AI",
      "release_date": "2025-12-08",
      "branch": "mixture_of_experts",
      "parents": [
        "mistral-large-2",
        "pixtral-large"
      ],
      "influences": [
        "llama-4-maverick",
        "deepseek-v3"
      ],
      "openness": "open_weights",
      "parameters": "675B total / 41B active",
      "context": "256K",
      "modalities": [
        "text",
        "image"
      ],
      "capabilities": [
        "tools",
        "multimodal",
        "open_weights",
        "long_context"
      ],
      "secondary_tags": [
        "open_weights",
        "multimodal",
        "frontier",
        "eu",
        "long_context"
      ],
      "notable": "Mistral's pivot to large MoE; matched closed frontier under Apache 2.0.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "mistral-small-4",
      "name": "Mistral Small 4",
      "vendor": "Mistral AI",
      "release_date": "2026-03-16",
      "branch": "mixture_of_experts",
      "parents": [
        "mistral-small-3",
        "magistral"
      ],
      "influences": [
        "mistral-large-3"
      ],
      "openness": "open_weights",
      "parameters": "119B total / 6B active",
      "context": "256K",
      "modalities": [
        "text",
        "image"
      ],
      "capabilities": [
        "tools",
        "multimodal",
        "open_weights",
        "long_context"
      ],
      "secondary_tags": [
        "open_weights",
        "hybrid_reasoning",
        "code_specialist",
        "eu",
        "long_context"
      ],
      "notable": "First widely-released hybrid mode-switching open-weight in the West.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "granite-4-0",
      "name": "IBM Granite 4.0",
      "vendor": "IBM",
      "release_date": "2025-10-02",
      "branch": "mixture_of_experts",
      "parents": [
        "granite-3-3"
      ],
      "influences": [
        "jamba",
        "falcon-mamba-7b"
      ],
      "openness": "open_weights",
      "parameters": "350M to 32B (e.g., H-Small 32B/9B active)",
      "context": "128K (eval) / 512K (train)",
      "modalities": [
        "text"
      ],
      "capabilities": [
        "open_weights",
        "long_context"
      ],
      "secondary_tags": [
        "open_weights",
        "ssm_hybrid",
        "mamba",
        "enterprise",
        "long_context"
      ],
      "notable": "70% lower memory and 2x faster inference than dense peers.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "falcon-mamba-7b",
      "name": "Falcon Mamba 7B",
      "vendor": "TII",
      "release_date": "2024-08-12",
      "branch": "decoder_only",
      "parents": [],
      "influences": [],
      "openness": "open_weights",
      "parameters": "7B",
      "context": "unbounded (constant-mem)",
      "modalities": [
        "text"
      ],
      "capabilities": [
        "open_weights"
      ],
      "secondary_tags": [
        "open_weights",
        "ssm_hybrid",
        "mamba",
        "attention_free",
        "uae"
      ],
      "notable": "First top-ranked pure SSM in the open.\n",
      "status": "active",
      "placement_confidence": "medium"
    },
    {
      "id": "flamingo",
      "name": "Flamingo",
      "vendor": "DeepMind",
      "release_date": "2022-04-29",
      "branch": "multimodal",
      "parents": [
        "chinchilla"
      ],
      "influences": [],
      "openness": "closed",
      "parameters": "3B / 9B / 80B",
      "context": "Variable",
      "modalities": [
        "text",
        "image",
        "video"
      ],
      "capabilities": [
        "multimodal"
      ],
      "secondary_tags": [
        "closed",
        "foundational"
      ],
      "notable": "Defined the few-shot VLM pattern; SOTA on 16 vision-language benchmarks; design precursor to Gemini's native multimodality.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "kosmos-1",
      "name": "Kosmos-1",
      "vendor": "Microsoft",
      "release_date": "2023-02-27",
      "branch": "multimodal",
      "parents": [],
      "influences": [
        "flamingo"
      ],
      "openness": "research_only",
      "parameters": "1.6B",
      "context": "2048",
      "modalities": [
        "text",
        "image"
      ],
      "capabilities": [
        "multimodal"
      ],
      "secondary_tags": [
        "research_only"
      ],
      "notable": "Early native MLLM from Microsoft.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "kosmos-2",
      "name": "Kosmos-2",
      "vendor": "Microsoft",
      "release_date": "2023-06-26",
      "branch": "multimodal",
      "parents": [
        "kosmos-1"
      ],
      "influences": [],
      "openness": "research_only",
      "parameters": "1.6B",
      "context": "2048",
      "modalities": [
        "text",
        "image"
      ],
      "capabilities": [
        "multimodal"
      ],
      "secondary_tags": [
        "research_only"
      ],
      "notable": "Pioneered bbox-as-tokens grounding pattern that influenced later VLMs.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "gpt-4v",
      "name": "GPT-4V (Vision)",
      "vendor": "OpenAI",
      "release_date": "2023-09-25",
      "branch": "multimodal",
      "parents": [
        "gpt-4"
      ],
      "influences": [],
      "openness": "closed",
      "parameters": "undisclosed",
      "context": "8K-128K",
      "modalities": [
        "text",
        "image"
      ],
      "capabilities": [
        "tools",
        "multimodal"
      ],
      "secondary_tags": [
        "closed"
      ],
      "notable": "First widely-deployed vision-capable LLM in production.\n",
      "status": "active",
      "placement_confidence": "medium"
    },
    {
      "id": "gpt-4o",
      "name": "GPT-4o",
      "vendor": "OpenAI",
      "release_date": "2024-05-13",
      "branch": "multimodal",
      "parents": [
        "gpt-4-turbo"
      ],
      "influences": [
        "gpt-4v"
      ],
      "openness": "closed",
      "parameters": "undisclosed",
      "context": "128K",
      "modalities": [
        "text",
        "image",
        "audio"
      ],
      "capabilities": [
        "tools",
        "multimodal",
        "long_context"
      ],
      "secondary_tags": [
        "closed",
        "historical_frontier",
        "long_context"
      ],
      "notable": "The \"omni\" moment. Free for ChatGPT users at launch.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "gpt-4o-mini",
      "name": "GPT-4o mini",
      "vendor": "OpenAI",
      "release_date": "2024-07-18",
      "branch": "multimodal",
      "parents": [
        "gpt-4o"
      ],
      "influences": [],
      "openness": "closed",
      "parameters": "undisclosed (small)",
      "context": "128K",
      "modalities": [
        "text",
        "image"
      ],
      "capabilities": [
        "tools",
        "multimodal",
        "long_context"
      ],
      "secondary_tags": [
        "closed",
        "small",
        "distilled",
        "long_context"
      ],
      "notable": "Cheaper than GPT-3.5 Turbo with significantly better quality.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "gpt-4-1",
      "name": "GPT-4.1",
      "vendor": "OpenAI",
      "release_date": "2025-04-14",
      "branch": "multimodal",
      "parents": [
        "gpt-4o"
      ],
      "influences": [
        "gpt-4-5"
      ],
      "openness": "closed",
      "parameters": "undisclosed",
      "context": "1M",
      "modalities": [
        "text",
        "image"
      ],
      "capabilities": [
        "tools",
        "multimodal",
        "long_context"
      ],
      "secondary_tags": [
        "closed",
        "code_specialist",
        "ultra_long_context"
      ],
      "notable": "First OpenAI flagship at 1M-token context.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "gpt-image-1",
      "name": "gpt-image-1",
      "vendor": "OpenAI",
      "release_date": "2025-04-23",
      "branch": "multimodal",
      "parents": [
        "gpt-4o"
      ],
      "influences": [],
      "openness": "closed",
      "parameters": "undisclosed",
      "context": "n/a",
      "modalities": [
        "text",
        "image"
      ],
      "capabilities": [
        "multimodal"
      ],
      "secondary_tags": [
        "closed"
      ],
      "notable": "DALL·E branding effectively retired in favor of this family.\n",
      "status": "active",
      "placement_confidence": "medium"
    },
    {
      "id": "sora",
      "name": "Sora (Sora Turbo)",
      "vendor": "OpenAI",
      "release_date": "2024-12-09",
      "branch": "multimodal",
      "parents": [],
      "influences": [
        "gpt-4"
      ],
      "openness": "closed",
      "parameters": "undisclosed",
      "context": "n/a (video)",
      "modalities": [
        "text",
        "video",
        "image"
      ],
      "capabilities": [
        "multimodal"
      ],
      "secondary_tags": [
        "closed",
        "video",
        "diffusion_transformer"
      ],
      "notable": "OpenAI's first major video-generation model. Architecturally a sibling, not a descendant, of the GPT line.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "sora-2",
      "name": "Sora 2",
      "vendor": "OpenAI",
      "release_date": "2025-09-30",
      "branch": "multimodal",
      "parents": [
        "sora"
      ],
      "influences": [],
      "openness": "closed",
      "parameters": "undisclosed",
      "context": "n/a (video)",
      "modalities": [
        "text",
        "video",
        "image",
        "audio"
      ],
      "capabilities": [
        "multimodal"
      ],
      "secondary_tags": [
        "closed",
        "video",
        "diffusion_transformer"
      ],
      "notable": "Closest OpenAI has come to a Veo-class video model.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "gemini-1",
      "name": "Gemini 1.0",
      "vendor": "Google + DeepMind",
      "release_date": "2023-12-06",
      "branch": "multimodal",
      "parents": [
        "palm-2"
      ],
      "influences": [
        "chinchilla",
        "flamingo"
      ],
      "openness": "closed",
      "parameters": "Nano (1.8B/3.25B) / Pro / Ultra (undisclosed)",
      "context": "32K",
      "modalities": [
        "text",
        "image",
        "audio",
        "video"
      ],
      "capabilities": [
        "multimodal"
      ],
      "secondary_tags": [
        "closed",
        "foundational"
      ],
      "notable": "First model to outperform human experts on MMLU (90.0% Ultra). Marked the merger of Google Brain plus DeepMind into one model line.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "gemini-1-5",
      "name": "Gemini 1.5 (Pro / Flash)",
      "vendor": "Google + DeepMind",
      "release_date": "2024-02-15",
      "branch": "multimodal",
      "parents": [
        "gemini-1"
      ],
      "influences": [
        "st-moe"
      ],
      "openness": "closed",
      "parameters": "Undisclosed; Pro and Flash tiers",
      "context": "1M (up to 2M experimentally)",
      "modalities": [
        "text",
        "image",
        "audio",
        "video"
      ],
      "capabilities": [
        "multimodal",
        "long_context"
      ],
      "secondary_tags": [
        "closed",
        "mixture_of_experts",
        "ultra_long_context"
      ],
      "notable": "First widely deployed 1M+ token context window.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "gemini-2",
      "name": "Gemini 2.0 (Flash / Pro / Flash-Lite)",
      "vendor": "Google + DeepMind",
      "release_date": "2024-12-11",
      "branch": "multimodal",
      "parents": [
        "gemini-1-5"
      ],
      "influences": [],
      "openness": "closed",
      "parameters": "Undisclosed (Flash / Pro / Flash-Lite tiers)",
      "context": "1M+",
      "modalities": [
        "text",
        "image",
        "audio",
        "video"
      ],
      "capabilities": [
        "tools",
        "multimodal",
        "agentic",
        "long_context"
      ],
      "secondary_tags": [
        "closed",
        "mixture_of_experts",
        "agentic",
        "long_context"
      ],
      "notable": "Agentic-era flagship. Native image-out replaces dedicated image gen models.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "gemma-3",
      "name": "Gemma 3",
      "vendor": "Google",
      "release_date": "2025-03-12",
      "branch": "multimodal",
      "parents": [
        "gemma-2"
      ],
      "influences": [
        "gemini-2"
      ],
      "openness": "open_weights",
      "parameters": "1B / 4B / 12B / 27B",
      "context": "128K",
      "modalities": [
        "text",
        "image"
      ],
      "capabilities": [
        "tools",
        "multimodal",
        "open_weights",
        "long_context"
      ],
      "secondary_tags": [
        "open_weights",
        "edge",
        "multilingual",
        "long_context"
      ],
      "notable": "Apache-2.0 licensed; fits on a single GPU/TPU.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "gemma-4",
      "name": "Gemma 4",
      "vendor": "Google",
      "release_date": "2026-03-31",
      "branch": "multimodal",
      "parents": [
        "gemma-3"
      ],
      "influences": [
        "gemini-3-pro"
      ],
      "openness": "open_weights",
      "parameters": "Multiple sizes (specifics not yet documented)",
      "context": "Not yet documented",
      "modalities": [
        "text",
        "image"
      ],
      "capabilities": [
        "open_weights",
        "multimodal"
      ],
      "secondary_tags": [
        "open_weights"
      ],
      "notable": "Latest Gemma sibling line; details still being published.\n",
      "status": "active",
      "placement_confidence": "medium"
    },
    {
      "id": "llama-3-2-vision",
      "name": "Llama 3.2 Vision (11B / 90B)",
      "vendor": "Meta AI",
      "release_date": "2024-09-25",
      "branch": "multimodal",
      "parents": [
        "llama-3-1"
      ],
      "influences": [],
      "openness": "open_weights",
      "parameters": "11B, 90B",
      "context": "128K",
      "modalities": [
        "text",
        "image"
      ],
      "capabilities": [
        "multimodal",
        "open_weights",
        "long_context"
      ],
      "secondary_tags": [
        "open_weights",
        "long_context"
      ],
      "notable": "Meta's first openly-released vision-language Llamas.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "pixtral-12b",
      "name": "Pixtral 12B",
      "vendor": "Mistral AI",
      "release_date": "2024-09-17",
      "branch": "multimodal",
      "parents": [
        "mistral-nemo-12b"
      ],
      "influences": [
        "llama-3-2-vision"
      ],
      "openness": "open_weights",
      "parameters": "12B + 400M vision encoder",
      "context": "128K",
      "modalities": [
        "text",
        "image"
      ],
      "capabilities": [
        "multimodal",
        "open_weights",
        "long_context"
      ],
      "secondary_tags": [
        "open_weights",
        "eu",
        "long_context"
      ],
      "notable": "Mistral's first multimodal model.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "pixtral-large",
      "name": "Pixtral Large (124B)",
      "vendor": "Mistral AI",
      "release_date": "2024-11-18",
      "branch": "multimodal",
      "parents": [
        "mistral-large-2"
      ],
      "influences": [
        "pixtral-12b"
      ],
      "openness": "open_weights",
      "parameters": "124B",
      "context": "128K",
      "modalities": [
        "text",
        "image"
      ],
      "capabilities": [
        "multimodal",
        "open_weights",
        "long_context"
      ],
      "secondary_tags": [
        "open_weights",
        "eu",
        "long_context"
      ],
      "notable": "Frontier open-weights vision model.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "voxtral",
      "name": "Voxtral 24B / 3B",
      "vendor": "Mistral AI",
      "release_date": "2025-07-15",
      "branch": "multimodal",
      "parents": [
        "mistral-small-3"
      ],
      "influences": [],
      "openness": "open_weights",
      "parameters": "24B, 3B",
      "context": "32K (up to 30-40 min audio)",
      "modalities": [
        "text",
        "audio"
      ],
      "capabilities": [
        "multimodal",
        "open_weights"
      ],
      "secondary_tags": [
        "open_weights",
        "audio",
        "eu"
      ],
      "notable": "Western open-weights speech-understanding frontier.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "qwen-2-5-vl",
      "name": "Qwen2.5-VL",
      "vendor": "Alibaba",
      "release_date": "2025-01",
      "branch": "multimodal",
      "parents": [
        "qwen-2-5"
      ],
      "influences": [],
      "openness": "open_weights",
      "parameters": "3B / 7B / 72B",
      "context": "128K",
      "modalities": [
        "text",
        "image",
        "video"
      ],
      "capabilities": [
        "multimodal",
        "open_weights",
        "long_context"
      ],
      "secondary_tags": [
        "open_weights",
        "video",
        "china",
        "long_context"
      ],
      "notable": "Strong open-weights vision-language baseline for the Chinese frontier.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "qwen-3-5",
      "name": "Qwen3.5",
      "vendor": "Alibaba",
      "release_date": "2026-02-24",
      "branch": "multimodal",
      "parents": [
        "qwen-3"
      ],
      "influences": [],
      "openness": "open_weights",
      "parameters": "Multiple sizes incl. ~397B reasoning variant",
      "context": "128K",
      "modalities": [
        "text",
        "image",
        "video"
      ],
      "capabilities": [
        "tools",
        "multimodal",
        "agentic",
        "open_weights",
        "long_context"
      ],
      "secondary_tags": [
        "open_weights",
        "mixture_of_experts",
        "agentic",
        "china",
        "long_context"
      ],
      "notable": "Bridge from Qwen3 to the 3.6 family; multilingual + multimodal scale.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "qwen-3-6-plus",
      "name": "Qwen3.6-Plus",
      "vendor": "Alibaba",
      "release_date": "2026-04-02",
      "branch": "multimodal",
      "parents": [
        "qwen-3-5"
      ],
      "influences": [],
      "openness": "closed",
      "parameters": "Undisclosed (flagship Plus)",
      "context": "1M",
      "modalities": [
        "text",
        "image",
        "video"
      ],
      "capabilities": [
        "tools",
        "multimodal",
        "agentic",
        "long_context"
      ],
      "secondary_tags": [
        "closed",
        "agentic",
        "frontier",
        "china",
        "ultra_long_context"
      ],
      "notable": "Alibaba's flagship agentic multimodal model.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "qwen-3-6-35b-a3b",
      "name": "Qwen3.6-35B-A3B",
      "vendor": "Alibaba",
      "release_date": "2026-04-16",
      "branch": "multimodal",
      "parents": [
        "qwen-3-6-plus"
      ],
      "influences": [],
      "openness": "open_weights",
      "parameters": "35B / 3B active",
      "context": "128K",
      "modalities": [
        "text",
        "image"
      ],
      "capabilities": [
        "tools",
        "multimodal",
        "agentic",
        "open_weights",
        "long_context"
      ],
      "secondary_tags": [
        "open_weights",
        "mixture_of_experts",
        "agentic",
        "china",
        "long_context"
      ],
      "notable": "73.4% SWE-bench Verified at 3B active parameters.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "doubao-1-5",
      "name": "Doubao 1.5 (Deep Thinking)",
      "vendor": "ByteDance",
      "release_date": "2025-04",
      "branch": "multimodal",
      "parents": [],
      "influences": [
        "deepseek-r1"
      ],
      "openness": "closed",
      "parameters": "20B total / 2B active",
      "context": "Undisclosed",
      "modalities": [
        "text",
        "image",
        "video"
      ],
      "capabilities": [
        "multimodal"
      ],
      "secondary_tags": [
        "closed",
        "mixture_of_experts",
        "video",
        "china"
      ],
      "notable": "Anchor of the Doubao multimodal line.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "doubao-1-6-vision",
      "name": "Doubao 1.6-Vision",
      "vendor": "ByteDance",
      "release_date": "2025-09",
      "branch": "multimodal",
      "parents": [
        "doubao-1-5"
      ],
      "influences": [],
      "openness": "closed",
      "parameters": "Undisclosed",
      "context": "Undisclosed",
      "modalities": [
        "text",
        "image"
      ],
      "capabilities": [
        "tools",
        "multimodal"
      ],
      "secondary_tags": [
        "closed",
        "tool_using",
        "china"
      ],
      "notable": "Bridge to the variable-thinking Doubao 1.6.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "ernie-4-5",
      "name": "ERNIE 4.5",
      "vendor": "Baidu",
      "release_date": "2025-03-16",
      "branch": "multimodal",
      "parents": [
        "ernie-4"
      ],
      "influences": [],
      "openness": "closed",
      "parameters": "Undisclosed",
      "context": "128K",
      "modalities": [
        "text",
        "image"
      ],
      "capabilities": [
        "multimodal",
        "long_context"
      ],
      "secondary_tags": [
        "closed",
        "china"
      ],
      "notable": "Anchor of the ERNIE multimodal line.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "ernie-5",
      "name": "ERNIE 5.0",
      "vendor": "Baidu",
      "release_date": "2026-01-22",
      "branch": "multimodal",
      "parents": [
        "ernie-4-5"
      ],
      "influences": [],
      "openness": "closed",
      "parameters": "2.4T (claimed)",
      "context": "Undisclosed",
      "modalities": [
        "text",
        "image",
        "audio",
        "video"
      ],
      "capabilities": [
        "multimodal"
      ],
      "secondary_tags": [
        "closed",
        "frontier",
        "china"
      ],
      "notable": "Preview Nov 2025; GA Jan 22, 2026; AIME 2025 87.0%.\n",
      "status": "active",
      "placement_confidence": "medium"
    },
    {
      "id": "minimax-vl-01",
      "name": "MiniMax-VL-01",
      "vendor": "MiniMax",
      "release_date": "2025-01-15",
      "branch": "multimodal",
      "parents": [
        "minimax-text-01"
      ],
      "influences": [],
      "openness": "open_weights",
      "parameters": "456B total / 45.9B active",
      "context": "1M",
      "modalities": [
        "text",
        "image"
      ],
      "capabilities": [
        "multimodal",
        "open_weights",
        "long_context"
      ],
      "secondary_tags": [
        "open_weights",
        "mixture_of_experts",
        "china",
        "ultra_long_context"
      ],
      "notable": "Carried Lightning Attention to multimodal at frontier scale.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "phi-4-multimodal",
      "name": "Phi-4-multimodal / Phi-4-mini",
      "vendor": "Microsoft",
      "release_date": "2025-02-26",
      "branch": "multimodal",
      "parents": [
        "phi-4"
      ],
      "influences": [],
      "openness": "open_weights",
      "parameters": "5.6B (multimodal); 3.8B (mini)",
      "context": "128K",
      "modalities": [
        "text",
        "image",
        "audio"
      ],
      "capabilities": [
        "multimodal",
        "open_weights",
        "long_context"
      ],
      "secondary_tags": [
        "open_weights",
        "small",
        "distilled",
        "audio",
        "long_context"
      ],
      "notable": "Phi family enters the unified omni-modal small-model space.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "falcon-2-11b",
      "name": "Falcon 2 11B (+ VLM)",
      "vendor": "TII",
      "release_date": "2024-05-13",
      "branch": "multimodal",
      "parents": [
        "falcon-180b"
      ],
      "influences": [],
      "openness": "open_weights",
      "parameters": "11B",
      "context": "8K",
      "modalities": [
        "text",
        "image"
      ],
      "capabilities": [
        "multimodal",
        "open_weights"
      ],
      "secondary_tags": [
        "open_weights",
        "uae"
      ],
      "notable": "TII's first multimodal release.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "granite-3-2",
      "name": "IBM Granite 3.2 (incl. Granite Vision 3.2 2B)",
      "vendor": "IBM",
      "release_date": "2025-02-26",
      "branch": "multimodal",
      "parents": [
        "granite-3-1"
      ],
      "influences": [],
      "openness": "open_weights",
      "parameters": "8B/2B Instruct; Vision 2B",
      "context": "128K",
      "modalities": [
        "text",
        "image"
      ],
      "capabilities": [
        "multimodal",
        "open_weights",
        "long_context"
      ],
      "secondary_tags": [
        "open_weights",
        "enterprise",
        "long_context"
      ],
      "notable": "First Granite multimodal; document-understanding focused vision model.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "o1-preview",
      "name": "OpenAI o1-preview",
      "vendor": "OpenAI",
      "release_date": "2024-09-12",
      "branch": "reasoning",
      "parents": [
        "gpt-4o"
      ],
      "influences": [
        "gpt-4-turbo"
      ],
      "openness": "closed",
      "parameters": "undisclosed",
      "context": "128K",
      "modalities": [
        "text"
      ],
      "capabilities": [
        "reasoning"
      ],
      "secondary_tags": [
        "closed",
        "frontier",
        "historical_frontier"
      ],
      "notable": "Launched the o-series. 83% on IMO qualifying exams vs. 13% for GPT-4o.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "o1",
      "name": "OpenAI o1",
      "vendor": "OpenAI",
      "release_date": "2024-12-05",
      "branch": "reasoning",
      "parents": [
        "o1-preview"
      ],
      "influences": [],
      "openness": "closed",
      "parameters": "undisclosed",
      "context": "128K-200K",
      "modalities": [
        "text",
        "image"
      ],
      "capabilities": [
        "tools",
        "multimodal",
        "reasoning"
      ],
      "secondary_tags": [
        "closed",
        "historical_frontier",
        "multimodal"
      ],
      "notable": "First reasoning model with full API support.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "o1-mini",
      "name": "OpenAI o1-mini",
      "vendor": "OpenAI",
      "release_date": "2024-09-12",
      "branch": "reasoning",
      "parents": [
        "o1-preview"
      ],
      "influences": [],
      "openness": "closed",
      "parameters": "undisclosed (small)",
      "context": "128K",
      "modalities": [
        "text"
      ],
      "capabilities": [
        "reasoning"
      ],
      "secondary_tags": [
        "closed",
        "small"
      ],
      "notable": "Made reasoning models accessible at low cost.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "o1-pro",
      "name": "OpenAI o1-pro",
      "vendor": "OpenAI",
      "release_date": "2024-12-05",
      "branch": "reasoning",
      "parents": [
        "o1"
      ],
      "influences": [],
      "openness": "closed",
      "parameters": "undisclosed",
      "context": "128K-200K",
      "modalities": [
        "text",
        "image"
      ],
      "capabilities": [
        "reasoning",
        "multimodal"
      ],
      "secondary_tags": [
        "closed",
        "multimodal"
      ],
      "notable": "Introduced the pro (extended-thinking) variant pattern.\n",
      "status": "active",
      "placement_confidence": "medium"
    },
    {
      "id": "o3-mini",
      "name": "OpenAI o3-mini",
      "vendor": "OpenAI",
      "release_date": "2025-01-31",
      "branch": "reasoning",
      "parents": [
        "o1-mini"
      ],
      "influences": [
        "o1"
      ],
      "openness": "closed",
      "parameters": "undisclosed (small)",
      "context": "200K",
      "modalities": [
        "text"
      ],
      "capabilities": [
        "tools",
        "reasoning"
      ],
      "secondary_tags": [
        "closed",
        "small",
        "tool_using"
      ],
      "notable": "First reasoning model offered to ChatGPT Free users.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "o3",
      "name": "OpenAI o3",
      "vendor": "OpenAI",
      "release_date": "2025-04-16",
      "branch": "reasoning",
      "parents": [
        "o1"
      ],
      "influences": [
        "o3-mini"
      ],
      "openness": "closed",
      "parameters": "undisclosed",
      "context": "200K",
      "modalities": [
        "text",
        "image"
      ],
      "capabilities": [
        "tools",
        "multimodal",
        "reasoning",
        "agentic"
      ],
      "secondary_tags": [
        "closed",
        "frontier",
        "agentic",
        "multimodal",
        "tool_using"
      ],
      "notable": "20% fewer major errors than o1.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "o4-mini",
      "name": "OpenAI o4-mini",
      "vendor": "OpenAI",
      "release_date": "2025-04-16",
      "branch": "reasoning",
      "parents": [
        "o3-mini"
      ],
      "influences": [
        "o3"
      ],
      "openness": "closed",
      "parameters": "undisclosed (small)",
      "context": "200K",
      "modalities": [
        "text",
        "image"
      ],
      "capabilities": [
        "tools",
        "multimodal",
        "reasoning",
        "agentic"
      ],
      "secondary_tags": [
        "closed",
        "small",
        "agentic",
        "multimodal",
        "tool_using"
      ],
      "notable": "99.5% pass@1 on AIME 2025 with tools.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "o3-pro",
      "name": "OpenAI o3-pro",
      "vendor": "OpenAI",
      "release_date": "2025-06-10",
      "branch": "reasoning",
      "parents": [
        "o3"
      ],
      "influences": [
        "o1-pro"
      ],
      "openness": "closed",
      "parameters": "undisclosed",
      "context": "200K",
      "modalities": [
        "text",
        "image"
      ],
      "capabilities": [
        "tools",
        "multimodal",
        "reasoning",
        "agentic"
      ],
      "secondary_tags": [
        "closed",
        "agentic",
        "multimodal",
        "tool_using"
      ],
      "notable": "Last pro variant before the GPT-5 line absorbed the extended-thinking pattern.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "gpt-5",
      "name": "GPT-5",
      "vendor": "OpenAI",
      "release_date": "2025-08-07",
      "branch": "reasoning",
      "parents": [
        "gpt-4-5",
        "o3"
      ],
      "influences": [
        "gpt-4o",
        "o4-mini"
      ],
      "openness": "closed",
      "parameters": "undisclosed",
      "context": "400K (128K max output)",
      "modalities": [
        "text",
        "image",
        "audio"
      ],
      "capabilities": [
        "tools",
        "mcp",
        "multimodal",
        "reasoning",
        "agentic",
        "long_context"
      ],
      "secondary_tags": [
        "closed",
        "frontier",
        "agentic",
        "multimodal",
        "mcp",
        "tool_using",
        "ultra_long_context"
      ],
      "notable": "Knit together OpenAI's two parallel lineages (pretraining-scaling and reasoning) into one product.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "gpt-5-codex",
      "name": "GPT-5-Codex",
      "vendor": "OpenAI",
      "release_date": "2025-09-15",
      "branch": "reasoning",
      "parents": [
        "gpt-5"
      ],
      "influences": [],
      "openness": "closed",
      "parameters": "undisclosed",
      "context": "400K",
      "modalities": [
        "text",
        "image"
      ],
      "capabilities": [
        "tools",
        "mcp",
        "multimodal",
        "reasoning",
        "agentic",
        "long_context"
      ],
      "secondary_tags": [
        "closed",
        "code_specialist",
        "agentic",
        "mcp",
        "tool_using"
      ],
      "notable": "Reuses the Codex brand from the 2021 GPT-3 code model but is a fundamentally different lineage.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "gpt-5-1",
      "name": "GPT-5.1 (Instant + Thinking + Pro)",
      "vendor": "OpenAI",
      "release_date": "2025-11-13",
      "branch": "reasoning",
      "parents": [
        "gpt-5"
      ],
      "influences": [],
      "openness": "closed",
      "parameters": "undisclosed",
      "context": "400K",
      "modalities": [
        "text",
        "image",
        "audio"
      ],
      "capabilities": [
        "tools",
        "mcp",
        "multimodal",
        "reasoning",
        "agentic",
        "long_context"
      ],
      "secondary_tags": [
        "closed",
        "frontier",
        "agentic",
        "multimodal",
        "ultra_long_context"
      ],
      "notable": "Established the Instant vs. Thinking UI split that all subsequent 5.x models inherit.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "gpt-5-2",
      "name": "GPT-5.2",
      "vendor": "OpenAI",
      "release_date": "2025-12-11",
      "branch": "reasoning",
      "parents": [
        "gpt-5-1"
      ],
      "influences": [],
      "openness": "closed",
      "parameters": "undisclosed",
      "context": "400K",
      "modalities": [
        "text",
        "image",
        "audio"
      ],
      "capabilities": [
        "tools",
        "mcp",
        "multimodal",
        "reasoning",
        "agentic",
        "long_context"
      ],
      "secondary_tags": [
        "closed",
        "frontier",
        "multimodal",
        "mcp",
        "ultra_long_context"
      ],
      "notable": "Last 5.x model to remain alongside 5.3 — Instant tier now uses 5.3, Thinking tier moved to 5.4.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "gpt-5-3-instant",
      "name": "GPT-5.3 Instant",
      "vendor": "OpenAI",
      "release_date": "2026-03-03",
      "branch": "reasoning",
      "parents": [
        "gpt-5-2"
      ],
      "influences": [],
      "openness": "closed",
      "parameters": "undisclosed",
      "context": "400K",
      "modalities": [
        "text",
        "image",
        "audio"
      ],
      "capabilities": [
        "tools",
        "mcp",
        "multimodal",
        "reasoning",
        "agentic",
        "long_context"
      ],
      "secondary_tags": [
        "closed",
        "multimodal",
        "mcp",
        "ultra_long_context"
      ],
      "notable": "OpenAI did NOT release a GPT-5.3 Thinking — the next thinking-tier release was GPT-5.4 Thinking.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "gpt-5-4",
      "name": "GPT-5.4 (Thinking + Pro + mini + nano)",
      "vendor": "OpenAI",
      "release_date": "2026-03-05",
      "branch": "reasoning",
      "parents": [
        "gpt-5-2",
        "gpt-5-codex"
      ],
      "influences": [
        "gpt-5-3-instant"
      ],
      "openness": "closed",
      "parameters": "undisclosed",
      "context": "400K+",
      "modalities": [
        "text",
        "image",
        "audio"
      ],
      "capabilities": [
        "tools",
        "mcp",
        "multimodal",
        "reasoning",
        "agentic",
        "long_context"
      ],
      "secondary_tags": [
        "closed",
        "frontier",
        "agentic",
        "multimodal",
        "code_specialist",
        "mcp",
        "ultra_long_context"
      ],
      "notable": "First explicit fusion of the GPT-5 base plus Codex training stacks into the main flagship.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "gpt-5-5",
      "name": "GPT-5.5",
      "vendor": "OpenAI",
      "release_date": "2026-04-23",
      "branch": "reasoning",
      "parents": [
        "gpt-5-4"
      ],
      "influences": [],
      "openness": "closed",
      "parameters": "undisclosed",
      "context": "400K+",
      "modalities": [
        "text",
        "image",
        "audio"
      ],
      "capabilities": [
        "tools",
        "mcp",
        "multimodal",
        "reasoning",
        "agentic",
        "long_context"
      ],
      "secondary_tags": [
        "closed",
        "agentic",
        "frontier",
        "multimodal",
        "ultra_long_context"
      ],
      "notable": "Released two days before this dataset (April 23, 2026). Terminal-Bench 2.0: 82.7%. GDPval (44 occupations): 84.9%. OSWorld-Verified: 78.7%.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "claude-3-7-sonnet",
      "name": "Claude 3.7 Sonnet",
      "vendor": "Anthropic",
      "release_date": "2025-02-24",
      "branch": "reasoning",
      "parents": [
        "claude-3-5-sonnet"
      ],
      "influences": [
        "o1"
      ],
      "openness": "closed",
      "parameters": "undisclosed",
      "context": "200K",
      "modalities": [
        "text",
        "image"
      ],
      "capabilities": [
        "tools",
        "multimodal",
        "reasoning",
        "long_context"
      ],
      "secondary_tags": [
        "closed",
        "multimodal",
        "hybrid_reasoning",
        "agentic",
        "tool_using",
        "long_context"
      ],
      "notable": "Anthropic's first hybrid reasoning model.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "claude-opus-4",
      "name": "Claude Opus 4",
      "vendor": "Anthropic",
      "release_date": "2025-05-22",
      "branch": "reasoning",
      "parents": [
        "claude-3-7-sonnet"
      ],
      "influences": [],
      "openness": "closed",
      "parameters": "undisclosed",
      "context": "200K",
      "modalities": [
        "text",
        "image"
      ],
      "capabilities": [
        "tools",
        "mcp",
        "multimodal",
        "reasoning",
        "agentic",
        "long_context"
      ],
      "secondary_tags": [
        "closed",
        "frontier",
        "multimodal",
        "extended_thinking",
        "agentic",
        "mcp",
        "tool_using",
        "long_context"
      ],
      "notable": "Top-tier coding model for long-running agentic tasks.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "claude-sonnet-4",
      "name": "Claude Sonnet 4",
      "vendor": "Anthropic",
      "release_date": "2025-05-22",
      "branch": "reasoning",
      "parents": [
        "claude-3-7-sonnet"
      ],
      "influences": [],
      "openness": "closed",
      "parameters": "undisclosed",
      "context": "200K (1M beta added Aug 12, 2025)",
      "modalities": [
        "text",
        "image"
      ],
      "capabilities": [
        "tools",
        "mcp",
        "multimodal",
        "reasoning",
        "long_context"
      ],
      "secondary_tags": [
        "closed",
        "multimodal",
        "extended_thinking",
        "mcp",
        "tool_using",
        "ultra_long_context"
      ],
      "notable": "First Claude to ship with 1M-token context (beta, Aug 2025).\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "claude-opus-4-1",
      "name": "Claude Opus 4.1",
      "vendor": "Anthropic",
      "release_date": "2025-08-05",
      "branch": "reasoning",
      "parents": [
        "claude-opus-4"
      ],
      "influences": [],
      "openness": "closed",
      "parameters": "undisclosed",
      "context": "200K",
      "modalities": [
        "text",
        "image"
      ],
      "capabilities": [
        "tools",
        "mcp",
        "multimodal",
        "reasoning",
        "agentic",
        "long_context"
      ],
      "secondary_tags": [
        "closed",
        "frontier",
        "multimodal",
        "extended_thinking",
        "agentic",
        "mcp",
        "tool_using",
        "long_context"
      ],
      "notable": "Improved coding and reasoning over Opus 4.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "claude-sonnet-4-5",
      "name": "Claude Sonnet 4.5",
      "vendor": "Anthropic",
      "release_date": "2025-09-29",
      "branch": "reasoning",
      "parents": [
        "claude-sonnet-4"
      ],
      "influences": [],
      "openness": "closed",
      "parameters": "undisclosed",
      "context": "200K",
      "modalities": [
        "text",
        "image"
      ],
      "capabilities": [
        "tools",
        "mcp",
        "multimodal",
        "reasoning",
        "agentic",
        "computer_use",
        "long_context"
      ],
      "secondary_tags": [
        "closed",
        "multimodal",
        "agentic",
        "extended_thinking",
        "computer_use",
        "mcp",
        "tool_using",
        "long_context"
      ],
      "notable": "Anthropic's flagship for agents/coding/computer-use through late 2025.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "claude-haiku-4-5",
      "name": "Claude Haiku 4.5",
      "vendor": "Anthropic",
      "release_date": "2025-10-15",
      "branch": "reasoning",
      "parents": [
        "claude-3-5-haiku"
      ],
      "influences": [
        "claude-sonnet-4"
      ],
      "openness": "closed",
      "parameters": "undisclosed",
      "context": "200K (64K output)",
      "modalities": [
        "text",
        "image"
      ],
      "capabilities": [
        "tools",
        "mcp",
        "multimodal",
        "reasoning",
        "agentic",
        "long_context"
      ],
      "secondary_tags": [
        "closed",
        "small",
        "multimodal",
        "extended_thinking",
        "mcp",
        "tool_using",
        "long_context"
      ],
      "notable": "First Haiku with extended-thinking. Skipped a Haiku 4 SKU.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "claude-opus-4-5",
      "name": "Claude Opus 4.5",
      "vendor": "Anthropic",
      "release_date": "2025-11-24",
      "branch": "reasoning",
      "parents": [
        "claude-opus-4-1"
      ],
      "influences": [],
      "openness": "closed",
      "parameters": "undisclosed",
      "context": "200K (1M beta added Feb 5, 2026)",
      "modalities": [
        "text",
        "image"
      ],
      "capabilities": [
        "tools",
        "mcp",
        "multimodal",
        "reasoning",
        "agentic",
        "computer_use",
        "long_context"
      ],
      "secondary_tags": [
        "closed",
        "frontier",
        "multimodal",
        "extended_thinking",
        "agentic",
        "computer_use",
        "mcp",
        "tool_using",
        "long_context"
      ],
      "notable": "80.9% on SWE-bench Verified at launch. Pricing dropped to $5/$25 per MTok — a third of prior Opus pricing.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "claude-opus-4-6",
      "name": "Claude Opus 4.6",
      "vendor": "Anthropic",
      "release_date": "2026-02-05",
      "branch": "reasoning",
      "parents": [
        "claude-opus-4-5"
      ],
      "influences": [],
      "openness": "closed",
      "parameters": "undisclosed",
      "context": "1M (GA March 13, 2026)",
      "modalities": [
        "text",
        "image"
      ],
      "capabilities": [
        "tools",
        "mcp",
        "multimodal",
        "reasoning",
        "agentic",
        "computer_use",
        "long_context"
      ],
      "secondary_tags": [
        "closed",
        "frontier",
        "multimodal",
        "adaptive_thinking",
        "agentic",
        "computer_use",
        "mcp",
        "tool_using",
        "ultra_long_context"
      ],
      "notable": "First Opus with 1M-token context. 76% MRCR-v2 retrieval at 1M tokens.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "claude-sonnet-4-6",
      "name": "Claude Sonnet 4.6",
      "vendor": "Anthropic",
      "release_date": "2026-02-17",
      "branch": "reasoning",
      "parents": [
        "claude-sonnet-4-5"
      ],
      "influences": [
        "claude-opus-4-6"
      ],
      "openness": "closed",
      "parameters": "undisclosed",
      "context": "1M (GA March 13, 2026)",
      "modalities": [
        "text",
        "image"
      ],
      "capabilities": [
        "tools",
        "mcp",
        "multimodal",
        "reasoning",
        "agentic",
        "computer_use",
        "long_context"
      ],
      "secondary_tags": [
        "closed",
        "multimodal",
        "extended_thinking",
        "agentic",
        "computer_use",
        "mcp",
        "tool_using",
        "ultra_long_context"
      ],
      "notable": "Sonnet-tier 1M-context model.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "claude-mythos",
      "name": "Claude Mythos Preview",
      "vendor": "Anthropic",
      "release_date": "2026-04-07",
      "branch": "reasoning",
      "parents": [
        "claude-opus-4-6"
      ],
      "influences": [],
      "openness": "gated",
      "parameters": "undisclosed",
      "context": "undisclosed",
      "modalities": [
        "text",
        "image"
      ],
      "capabilities": [
        "tools",
        "multimodal",
        "reasoning",
        "agentic"
      ],
      "secondary_tags": [
        "gated",
        "frontier",
        "multimodal",
        "agentic"
      ],
      "notable": "Invitation-only research preview as part of Project Glasswing. 83.1% on CyberGym vs. Opus 4.6's ~74%. NOT generally available.\n",
      "status": "gated",
      "placement_confidence": "medium"
    },
    {
      "id": "claude-opus-4-7",
      "name": "Claude Opus 4.7",
      "vendor": "Anthropic",
      "release_date": "2026-04-16",
      "branch": "reasoning",
      "parents": [
        "claude-opus-4-6"
      ],
      "influences": [
        "claude-mythos"
      ],
      "openness": "closed",
      "parameters": "undisclosed",
      "context": "1M",
      "modalities": [
        "text",
        "image"
      ],
      "capabilities": [
        "tools",
        "mcp",
        "multimodal",
        "reasoning",
        "agentic",
        "computer_use",
        "long_context"
      ],
      "secondary_tags": [
        "closed",
        "frontier",
        "multimodal",
        "adaptive_thinking",
        "agentic",
        "computer_use",
        "mcp",
        "tool_using",
        "ultra_long_context"
      ],
      "notable": "Most capable generally available Claude as of April 25, 2026 (Mythos Preview is more capable but gated).\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "claude-opus-4-8",
      "name": "Claude Opus 4.8",
      "vendor": "Anthropic",
      "release_date": "2026-05-28",
      "branch": "reasoning",
      "parents": [
        "claude-opus-4-7"
      ],
      "influences": [
        "claude-mythos"
      ],
      "openness": "closed",
      "parameters": "undisclosed",
      "context": "1M",
      "modalities": [
        "text",
        "image"
      ],
      "capabilities": [
        "tools",
        "mcp",
        "multimodal",
        "reasoning",
        "agentic",
        "computer_use",
        "long_context"
      ],
      "secondary_tags": [
        "closed",
        "frontier",
        "multimodal",
        "adaptive_thinking",
        "agentic",
        "computer_use",
        "mcp",
        "tool_using",
        "ultra_long_context"
      ],
      "notable": "Retook the #1 spot on the Artificial Analysis Intelligence Index (61.4 vs GPT-5.5's 60.2) on May 28, 2026, with SWE-Bench Pro 69.2%; headline pricing held at $5/$25 per Mtok while fast mode dropped ~3x.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "gemini-2-5-pro",
      "name": "Gemini 2.5 Pro",
      "vendor": "Google + DeepMind",
      "release_date": "2025-03-25",
      "branch": "reasoning",
      "parents": [
        "gemini-2"
      ],
      "influences": [],
      "openness": "closed",
      "parameters": "Undisclosed (sparse MoE)",
      "context": "1M",
      "modalities": [
        "text",
        "image",
        "audio",
        "video"
      ],
      "capabilities": [
        "tools",
        "multimodal",
        "reasoning",
        "agentic",
        "long_context"
      ],
      "secondary_tags": [
        "closed",
        "mixture_of_experts",
        "multimodal",
        "ultra_long_context"
      ],
      "notable": "First Gemini thinking model; debuted #1 on LMArena.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "gemini-3-pro",
      "name": "Gemini 3 Pro",
      "vendor": "Google + DeepMind",
      "release_date": "2025-11-18",
      "branch": "reasoning",
      "parents": [
        "gemini-2-5-pro"
      ],
      "influences": [],
      "openness": "closed",
      "parameters": "Undisclosed",
      "context": "1M+",
      "modalities": [
        "text",
        "image",
        "audio",
        "video"
      ],
      "capabilities": [
        "tools",
        "multimodal",
        "reasoning",
        "agentic",
        "long_context"
      ],
      "secondary_tags": [
        "closed",
        "mixture_of_experts",
        "multimodal",
        "agentic",
        "frontier",
        "ultra_long_context"
      ],
      "notable": "91.9% GPQA Diamond, 23.4% MathArena Apex, 81% MMMU-Pro.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "gemini-3-1-pro",
      "name": "Gemini 3.1 Pro",
      "vendor": "Google + DeepMind",
      "release_date": "2026-02-19",
      "branch": "reasoning",
      "parents": [
        "gemini-3-pro"
      ],
      "influences": [],
      "openness": "closed",
      "parameters": "Undisclosed",
      "context": "1M+",
      "modalities": [
        "text",
        "image",
        "audio",
        "video"
      ],
      "capabilities": [
        "tools",
        "multimodal",
        "reasoning",
        "agentic",
        "long_context"
      ],
      "secondary_tags": [
        "closed",
        "multimodal",
        "agentic",
        "frontier",
        "ultra_long_context"
      ],
      "notable": "Underpins Google's April 2026 AI research agents product.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "deepseek-r1",
      "name": "DeepSeek-R1",
      "vendor": "DeepSeek AI",
      "release_date": "2025-01-20",
      "branch": "reasoning",
      "parents": [
        "deepseek-v3"
      ],
      "influences": [
        "o1"
      ],
      "openness": "open_weights",
      "parameters": "671B total / 37B active",
      "context": "128K",
      "modalities": [
        "text"
      ],
      "capabilities": [
        "reasoning",
        "open_weights",
        "long_context"
      ],
      "secondary_tags": [
        "open_weights",
        "mixture_of_experts",
        "frontier",
        "china",
        "long_context"
      ],
      "notable": "Open-weights reasoning model that triggered the 2025 efficiency-shock cycle.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "deepseek-r1-distill",
      "name": "DeepSeek-R1-Distill (Qwen / Llama family)",
      "vendor": "DeepSeek AI",
      "release_date": "2025-01-22",
      "branch": "reasoning",
      "parents": [
        "deepseek-r1"
      ],
      "influences": [
        "qwen-2-5",
        "llama-3"
      ],
      "openness": "open_weights",
      "parameters": "1.5B / 7B / 8B / 14B / 32B / 70B",
      "context": "32K-128K",
      "modalities": [
        "text"
      ],
      "capabilities": [
        "reasoning",
        "open_weights"
      ],
      "secondary_tags": [
        "open_weights",
        "distilled",
        "china"
      ],
      "notable": "Made R1-class reasoning available in small open-weights sizes.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "deepseek-r2",
      "name": "DeepSeek-R2",
      "vendor": "DeepSeek AI",
      "release_date": "2026-04",
      "branch": "reasoning",
      "parents": [
        "deepseek-r1"
      ],
      "influences": [
        "deepseek-v3-2"
      ],
      "openness": "open_weights",
      "parameters": "32B (dense)",
      "context": "128K",
      "modalities": [
        "text"
      ],
      "capabilities": [
        "reasoning",
        "open_weights",
        "long_context"
      ],
      "secondary_tags": [
        "open_weights",
        "frontier",
        "china",
        "long_context"
      ],
      "notable": "Strategic shift to consumer-GPU-runnable reasoning; 92.7% AIME 2025.\n",
      "status": "active",
      "placement_confidence": "medium"
    },
    {
      "id": "qwq-32b-preview",
      "name": "QwQ-32B-Preview",
      "vendor": "Alibaba",
      "release_date": "2024-11",
      "branch": "reasoning",
      "parents": [
        "qwen-2-5"
      ],
      "influences": [
        "o1"
      ],
      "openness": "open_weights",
      "parameters": "32B",
      "context": "32K",
      "modalities": [
        "text"
      ],
      "capabilities": [
        "reasoning",
        "open_weights"
      ],
      "secondary_tags": [
        "open_weights",
        "china"
      ],
      "notable": "Alibaba's first reasoning-model entry.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "qwen-3-thinking-2507",
      "name": "Qwen3-235B-A22B-Thinking-2507",
      "vendor": "Alibaba",
      "release_date": "2025-07",
      "branch": "reasoning",
      "parents": [
        "qwen-3"
      ],
      "influences": [],
      "openness": "open_weights",
      "parameters": "235B / 22B active",
      "context": "128K",
      "modalities": [
        "text"
      ],
      "capabilities": [
        "reasoning",
        "open_weights",
        "long_context"
      ],
      "secondary_tags": [
        "open_weights",
        "mixture_of_experts",
        "china",
        "long_context"
      ],
      "notable": "Qwen3 reasoning-mode flagship.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "kimi-k1-5",
      "name": "Kimi K1.5",
      "vendor": "Moonshot AI",
      "release_date": "2025-01-20",
      "branch": "reasoning",
      "parents": [
        "kimi-k1"
      ],
      "influences": [
        "o1"
      ],
      "openness": "closed",
      "parameters": "Undisclosed",
      "context": "128K",
      "modalities": [
        "text",
        "image"
      ],
      "capabilities": [
        "reasoning",
        "multimodal",
        "long_context"
      ],
      "secondary_tags": [
        "closed",
        "multimodal",
        "china",
        "long_context"
      ],
      "notable": "First Kimi reasoning model.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "kimi-k2-5",
      "name": "Kimi K2.5",
      "vendor": "Moonshot AI",
      "release_date": "2026-01-26",
      "branch": "reasoning",
      "parents": [
        "kimi-k2"
      ],
      "influences": [],
      "openness": "open_weights",
      "parameters": "1T total / 32B active",
      "context": "256K",
      "modalities": [
        "text",
        "image"
      ],
      "capabilities": [
        "tools",
        "reasoning",
        "multimodal",
        "agentic",
        "open_weights",
        "long_context"
      ],
      "secondary_tags": [
        "open_weights",
        "mixture_of_experts",
        "multimodal",
        "agentic",
        "china",
        "long_context"
      ],
      "notable": "Bridge from Kimi K2 to the K2.6 multi-agent flagship.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "glm-z1",
      "name": "GLM-Z1-32B-0414",
      "vendor": "Z.AI (Zhipu)",
      "release_date": "2025-04",
      "branch": "reasoning",
      "parents": [
        "glm-4-32b-0414"
      ],
      "influences": [
        "deepseek-r1"
      ],
      "openness": "open_weights",
      "parameters": "32B (also 9B variant)",
      "context": "128K",
      "modalities": [
        "text"
      ],
      "capabilities": [
        "tools",
        "reasoning",
        "open_weights",
        "long_context"
      ],
      "secondary_tags": [
        "open_weights",
        "china",
        "long_context"
      ],
      "notable": "Zhipu's open-weights reasoning entry.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "glm-5-1",
      "name": "GLM-5.1",
      "vendor": "Z.AI (Zhipu)",
      "release_date": "2026-04-08",
      "branch": "reasoning",
      "parents": [
        "glm-5"
      ],
      "influences": [],
      "openness": "open_weights",
      "parameters": "754B total",
      "context": "203K",
      "modalities": [
        "text"
      ],
      "capabilities": [
        "tools",
        "reasoning",
        "agentic",
        "multi_agent_native",
        "open_weights",
        "long_context"
      ],
      "secondary_tags": [
        "open_weights",
        "mixture_of_experts",
        "agentic",
        "multi_agent_native",
        "frontier",
        "china",
        "long_context"
      ],
      "notable": "SOTA SWE-Bench Pro 58.4 — beats GPT-5.4, Claude Opus 4.6, Gemini 3.1 Pro.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "doubao-1-6",
      "name": "Doubao 1.6",
      "vendor": "ByteDance",
      "release_date": "2025-10",
      "branch": "reasoning",
      "parents": [
        "doubao-1-6-vision"
      ],
      "influences": [],
      "openness": "closed",
      "parameters": "Undisclosed",
      "context": "Undisclosed",
      "modalities": [
        "text",
        "image"
      ],
      "capabilities": [
        "reasoning",
        "multimodal"
      ],
      "secondary_tags": [
        "closed",
        "mixture_of_experts",
        "multimodal",
        "adaptive_thinking",
        "china"
      ],
      "notable": "First Chinese adaptive-thinking model.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "hunyuan-t1",
      "name": "Hunyuan T1",
      "vendor": "Tencent",
      "release_date": "2025-03-21",
      "branch": "reasoning",
      "parents": [
        "hunyuan-turbos"
      ],
      "influences": [
        "deepseek-r1"
      ],
      "openness": "closed",
      "parameters": "Undisclosed (MoE)",
      "context": "128K",
      "modalities": [
        "text"
      ],
      "capabilities": [
        "reasoning",
        "long_context"
      ],
      "secondary_tags": [
        "closed",
        "ssm_hybrid",
        "mamba",
        "china",
        "long_context"
      ],
      "notable": "First ultra-large SSM-hybrid reasoning model.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "minimax-m1",
      "name": "MiniMax-M1",
      "vendor": "MiniMax",
      "release_date": "2025-06-16",
      "branch": "reasoning",
      "parents": [
        "minimax-text-01"
      ],
      "influences": [
        "deepseek-r1"
      ],
      "openness": "open_weights",
      "parameters": "456B total / 45.9B active",
      "context": "1M (80K reasoning output)",
      "modalities": [
        "text"
      ],
      "capabilities": [
        "reasoning",
        "open_weights",
        "long_context"
      ],
      "secondary_tags": [
        "open_weights",
        "mixture_of_experts",
        "china",
        "ultra_long_context"
      ],
      "notable": "First open-weights Lightning-Attention reasoning model.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "ernie-x1",
      "name": "ERNIE X1",
      "vendor": "Baidu",
      "release_date": "2025-03-16",
      "branch": "reasoning",
      "parents": [
        "ernie-4-5"
      ],
      "influences": [
        "deepseek-r1"
      ],
      "openness": "closed",
      "parameters": "Undisclosed",
      "context": "128K",
      "modalities": [
        "text",
        "image"
      ],
      "capabilities": [
        "tools",
        "reasoning",
        "multimodal"
      ],
      "secondary_tags": [
        "closed",
        "multimodal",
        "china"
      ],
      "notable": "Baidu's reasoning-model entry.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "grok-3",
      "name": "Grok-3 (incl. Think / Big Brain)",
      "vendor": "xAI",
      "release_date": "2025-02-18",
      "branch": "reasoning",
      "parents": [
        "grok-2"
      ],
      "influences": [
        "o1"
      ],
      "openness": "closed",
      "parameters": "Undisclosed",
      "context": "1M",
      "modalities": [
        "text"
      ],
      "capabilities": [
        "reasoning",
        "long_context"
      ],
      "secondary_tags": [
        "closed",
        "frontier",
        "ultra_long_context"
      ],
      "notable": "xAI's first reasoning model.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "grok-4",
      "name": "Grok 4",
      "vendor": "xAI",
      "release_date": "2025-07-09",
      "branch": "reasoning",
      "parents": [
        "grok-3"
      ],
      "influences": [],
      "openness": "closed",
      "parameters": "Undisclosed",
      "context": "256K",
      "modalities": [
        "text"
      ],
      "capabilities": [
        "tools",
        "reasoning",
        "agentic",
        "long_context"
      ],
      "secondary_tags": [
        "closed",
        "agentic",
        "frontier",
        "tool_using",
        "long_context"
      ],
      "notable": "Most agentic Grok at release.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "grok-4-20",
      "name": "Grok 4.20",
      "vendor": "xAI",
      "release_date": "2026-02-17",
      "branch": "reasoning",
      "parents": [
        "grok-4"
      ],
      "influences": [],
      "openness": "closed",
      "parameters": "Undisclosed",
      "context": "2M",
      "modalities": [
        "text"
      ],
      "capabilities": [
        "tools",
        "reasoning",
        "agentic",
        "multi_agent_native",
        "long_context"
      ],
      "secondary_tags": [
        "closed",
        "agentic",
        "multi_agent_native",
        "frontier",
        "ultra_long_context"
      ],
      "notable": "8x context vs Grok-4; Chatbot Arena ELO ~1493.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "magistral",
      "name": "Magistral",
      "vendor": "Mistral AI",
      "release_date": "2025-06-10",
      "branch": "reasoning",
      "parents": [
        "mistral-small-3"
      ],
      "influences": [
        "deepseek-r1",
        "o1"
      ],
      "openness": "open_weights",
      "parameters": "Small 24B (open); Medium (closed preview)",
      "context": "40K",
      "modalities": [
        "text"
      ],
      "capabilities": [
        "reasoning",
        "open_weights"
      ],
      "secondary_tags": [
        "open_weights",
        "multilingual",
        "eu"
      ],
      "notable": "First fully Western open-weights reasoning model after DeepSeek-R1.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "phi-4-reasoning",
      "name": "Phi-4-reasoning / mini-reasoning",
      "vendor": "Microsoft",
      "release_date": "2025-04-30",
      "branch": "reasoning",
      "parents": [
        "phi-4"
      ],
      "influences": [
        "o1",
        "deepseek-r1"
      ],
      "openness": "open_weights",
      "parameters": "14B (reasoning/+); 3.8B (mini-reasoning); 15B (reasoning-vision)",
      "context": "32K-128K",
      "modalities": [
        "text",
        "image"
      ],
      "capabilities": [
        "reasoning",
        "open_weights",
        "long_context"
      ],
      "secondary_tags": [
        "open_weights",
        "small",
        "distilled",
        "long_context"
      ],
      "notable": "Beat OpenAI o1-mini and DeepSeek-R1-Distill-Llama-70B at 14B.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "mimo-v2-5-pro",
      "name": "MiMo V2.5 Pro",
      "vendor": "Xiaomi",
      "release_date": "2026-04-27",
      "branch": "mixture_of_experts",
      "parents": [],
      "influences": [
        "kimi-k2-6",
        "deepseek-v4-pro"
      ],
      "openness": "open_weights",
      "parameters": "1.02T total / 42B active",
      "context": "1M",
      "modalities": [
        "text"
      ],
      "capabilities": [
        "tools",
        "reasoning",
        "agentic",
        "open_weights",
        "long_context"
      ],
      "secondary_tags": [
        "open_weights",
        "mixture_of_experts",
        "agentic",
        "frontier",
        "china",
        "ultra_long_context"
      ],
      "notable": "Ties Kimi K2.6 at #1 open-weights on Artificial Analysis Intelligence Index (54); Xiaomi shipped the 100T-free-token Orbit Plan alongside the weights to anchor evaluation.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "nemotron-3-nano-omni",
      "name": "Nemotron 3 Nano Omni",
      "vendor": "NVIDIA",
      "release_date": "2026-04-28",
      "branch": "multimodal",
      "parents": [],
      "influences": [],
      "openness": "open_weights",
      "parameters": "30B total / 3B active",
      "context": "256K",
      "modalities": [
        "text",
        "image",
        "video",
        "audio"
      ],
      "capabilities": [
        "multimodal",
        "reasoning",
        "agentic",
        "open_weights",
        "long_context"
      ],
      "secondary_tags": [
        "open_weights",
        "mixture_of_experts",
        "multimodal",
        "edge",
        "agentic",
        "ssm_hybrid"
      ],
      "notable": "First-party NVIDIA open-weights multimodal MoE — the only Western entry in the open-frontier multimodal slot this week.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "openai-privacy-filter",
      "name": "OpenAI Privacy Filter",
      "vendor": "OpenAI",
      "release_date": "2026-04-28",
      "branch": "encoder_only",
      "parents": [],
      "influences": [],
      "openness": "open_weights",
      "parameters": "1.5B total / 50M active",
      "context": "128K",
      "modalities": [
        "text"
      ],
      "capabilities": [
        "open_weights"
      ],
      "secondary_tags": [
        "open_weights",
        "mixture_of_experts",
        "small",
        "edge"
      ],
      "notable": "OpenAI's first true open-weights model since GPT-2 — signals that narrow security primitives can ship openly while frontier models stay closed.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "laguna-m-1",
      "name": "Laguna M.1",
      "vendor": "Poolside",
      "release_date": "2026-04-28",
      "branch": "mixture_of_experts",
      "parents": [],
      "influences": [],
      "openness": "closed",
      "parameters": "225B total / 23B active",
      "context": "long",
      "modalities": [
        "text"
      ],
      "capabilities": [
        "tools",
        "agentic"
      ],
      "secondary_tags": [
        "closed",
        "mixture_of_experts",
        "agentic",
        "code_specialist",
        "us"
      ],
      "notable": "46.9 SWE-Bench Pro / 40.7 Terminal-Bench 2.0; shipped alongside the 'pool' terminal agent and 'Shimmer' cloud IDE — the unit of competition is now agent runtime, not weights alone.\n",
      "status": "active",
      "placement_confidence": "medium"
    },
    {
      "id": "laguna-xs-2",
      "name": "Laguna XS.2",
      "vendor": "Poolside",
      "release_date": "2026-04-28",
      "branch": "mixture_of_experts",
      "parents": [
        "laguna-m-1"
      ],
      "influences": [],
      "openness": "open_weights",
      "parameters": "33B total / 3B active",
      "context": "long",
      "modalities": [
        "text"
      ],
      "capabilities": [
        "tools",
        "agentic",
        "open_weights"
      ],
      "secondary_tags": [
        "open_weights",
        "mixture_of_experts",
        "agentic",
        "code_specialist",
        "edge",
        "small",
        "us"
      ],
      "notable": "44.5 SWE-Bench Pro at one-eighth the active params of M.1 — narrowest open-weights coding gap to a flagship sibling currently on file.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "grok-4-3",
      "name": "Grok 4.3",
      "vendor": "xAI",
      "release_date": "2026-04-30",
      "branch": "reasoning",
      "parents": [
        "grok-4-20"
      ],
      "influences": [],
      "openness": "closed",
      "parameters": "Undisclosed",
      "context": "1M",
      "modalities": [
        "text"
      ],
      "capabilities": [
        "tools",
        "agentic",
        "long_context"
      ],
      "secondary_tags": [
        "closed",
        "agentic",
        "frontier",
        "tool_using",
        "ultra_long_context"
      ],
      "notable": "Priced at $1.25/$2.50 per M tokens with 207 tok/s output — cost-efficient agent runtime, not frontier intelligence (AA Index 53 vs 60 for GPT-5.5).\n",
      "status": "active",
      "placement_confidence": "medium"
    },
    {
      "id": "gpt-5-5-instant",
      "name": "GPT-5.5 Instant",
      "vendor": "OpenAI",
      "release_date": "2026-05-05",
      "branch": "reasoning",
      "parents": [
        "gpt-5-5"
      ],
      "influences": [],
      "openness": "closed",
      "parameters": "Undisclosed",
      "context": "Undisclosed",
      "modalities": [
        "text",
        "image"
      ],
      "capabilities": [
        "reasoning",
        "tools",
        "mcp",
        "multimodal"
      ],
      "secondary_tags": [
        "closed",
        "frontier",
        "multimodal",
        "tool_using",
        "mcp"
      ],
      "notable": "52.5% fewer hallucinations on high-stakes prompts vs GPT-5.3 Instant; AIME 2025 81.2 (vs 65.4); MMMU-Pro 76.0. The 'cheap fallback' tier just inherited frontier-grade math.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "gpt-realtime-2",
      "name": "GPT-Realtime-2",
      "vendor": "OpenAI",
      "release_date": "2026-05-07",
      "branch": "multimodal",
      "parents": [],
      "influences": [
        "gpt-5-5"
      ],
      "openness": "closed",
      "parameters": "Undisclosed",
      "context": "128K",
      "modalities": [
        "text",
        "audio"
      ],
      "capabilities": [
        "multimodal",
        "reasoning",
        "tools",
        "agentic",
        "long_context"
      ],
      "secondary_tags": [
        "closed",
        "frontier",
        "audio",
        "agentic",
        "tool_using",
        "long_context"
      ],
      "notable": "Big Bench Audio 96.6% (high effort, +15.2pp vs v1.5); Audio MultiChallenge 48.5% xhigh (+13.8pp). Voice crossed into chain-of-thought territory — 'voice model' is now its own tier with Big Bench Audio / Speech Arena leaderboards.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "gpt-realtime-translate",
      "name": "GPT-Realtime-Translate",
      "vendor": "OpenAI",
      "release_date": "2026-05-07",
      "branch": "multimodal",
      "parents": [
        "gpt-realtime-2"
      ],
      "influences": [],
      "openness": "closed",
      "parameters": "Undisclosed",
      "context": "Undisclosed",
      "modalities": [
        "text",
        "audio"
      ],
      "capabilities": [
        "multimodal"
      ],
      "secondary_tags": [
        "closed",
        "audio",
        "multilingual"
      ],
      "notable": "BolnaAI independent eval shows 12.5% lower WER on Hindi / Tamil / Telugu vs prior generation.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "gpt-realtime-whisper",
      "name": "GPT-Realtime-Whisper",
      "vendor": "OpenAI",
      "release_date": "2026-05-07",
      "branch": "multimodal",
      "parents": [],
      "influences": [
        "gpt-realtime-2"
      ],
      "openness": "closed",
      "parameters": "Undisclosed",
      "context": "Undisclosed",
      "modalities": [
        "text",
        "audio"
      ],
      "capabilities": [
        "multimodal"
      ],
      "secondary_tags": [
        "closed",
        "audio",
        "edge"
      ],
      "notable": "Streaming-native STT at half the per-minute price of the Translate sibling — completes the OpenAI voice trio for developers building real-time voice agents.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "inworld-realtime-tts-2",
      "name": "Inworld Realtime TTS-2",
      "vendor": "Inworld",
      "release_date": "2026-05-05",
      "branch": "multimodal",
      "parents": [],
      "influences": [],
      "openness": "closed",
      "parameters": "Undisclosed",
      "context": "Undisclosed",
      "modalities": [
        "text",
        "audio"
      ],
      "capabilities": [
        "multimodal"
      ],
      "secondary_tags": [
        "closed",
        "audio",
        "multilingual",
        "agentic"
      ],
      "notable": "Inworld TTS line holds #1 on Artificial Analysis Speech Arena (Elo 1209.6, 73.3% win rate). TTS-2 extends into real-time dialog — the second 'frontier voice' release of the same week as the OpenAI Realtime trio.\n",
      "status": "active",
      "placement_confidence": "medium"
    },
    {
      "id": "zaya1-8b",
      "name": "ZAYA1-8B",
      "vendor": "Zyphra",
      "release_date": "2026-05-06",
      "branch": "reasoning",
      "parents": [],
      "influences": [
        "deepseek-v4-pro",
        "mimo-v2-5-pro"
      ],
      "openness": "open_weights",
      "parameters": "8B total / 700M active",
      "context": "32K",
      "modalities": [
        "text"
      ],
      "capabilities": [
        "reasoning",
        "open_weights"
      ],
      "secondary_tags": [
        "open_weights",
        "mixture_of_experts",
        "frontier",
        "small",
        "edge"
      ],
      "notable": "AIME 2025 91.9% / HMMT 2025 89.6% with Markovian RSA test-time compute; matches DeepSeek-R1-0528 single-rollout at <1B active — and proves AMD is now a credible frontier-training substrate.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "emo-1b14b",
      "name": "EMO (Emergent Mixture-of-Experts)",
      "vendor": "Allen Institute for AI",
      "release_date": "2026-05-08",
      "branch": "mixture_of_experts",
      "parents": [],
      "influences": [],
      "openness": "open_source",
      "parameters": "14B total / 1B active (8 of 128 experts)",
      "context": "Undisclosed",
      "modalities": [
        "text"
      ],
      "capabilities": [
        "open_weights"
      ],
      "secondary_tags": [
        "open_source",
        "research_only",
        "small"
      ],
      "notable": "Maintains near-full performance with only 12.5% of experts active (3pp drop) — the line of research that will set the next 12 months of MoE serving economics.\n",
      "status": "active",
      "placement_confidence": "medium"
    },
    {
      "id": "minicpm-v-4-6",
      "name": "MiniCPM-V 4.6 1.3B",
      "vendor": "OpenBMB",
      "release_date": "2026-05-11",
      "branch": "multimodal",
      "parents": [],
      "influences": [],
      "openness": "open_weights",
      "parameters": "1.3B (SigLIP2-400M vision + Qwen3.5-0.8B LLM backbone)",
      "context": "262K",
      "modalities": [
        "text",
        "image",
        "video"
      ],
      "capabilities": [
        "multimodal",
        "long_context",
        "open_weights"
      ],
      "secondary_tags": [
        "open_weights",
        "multimodal",
        "edge",
        "small",
        "china"
      ],
      "notable": "Hits AA Intelligence Index 13 — beats Qwen3.5-0.8B at 19x lower token cost; ships native deployment for iOS / Android / HarmonyOS, collapsing the friction between open-weights release and phone build.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "perceptron-mk1",
      "name": "Perceptron Mk1",
      "vendor": "Perceptron",
      "release_date": "2026-05-12",
      "branch": "multimodal",
      "parents": [],
      "influences": [],
      "openness": "closed",
      "parameters": "Undisclosed",
      "context": "32K (multimodal)",
      "modalities": [
        "text",
        "image",
        "video"
      ],
      "capabilities": [
        "multimodal",
        "reasoning",
        "tools",
        "agentic"
      ],
      "secondary_tags": [
        "closed",
        "multimodal",
        "frontier",
        "video",
        "us"
      ],
      "notable": "Matches Gemini Pro on video / embodied benchmarks at $0.15 / $1.50 per Mtok — cheaper than Gemini Flash Lite; 85.1 EmbSpatialBench, 72.4 RefSpatialBench. Pricing collapse for physical-AI perception.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "sana-wm",
      "name": "SANA-WM",
      "vendor": "NVIDIA",
      "release_date": "2026-05-15",
      "branch": "multimodal",
      "parents": [],
      "influences": [],
      "openness": "open_weights",
      "parameters": "2.6B (Hybrid Linear Diffusion Transformer)",
      "context": "60 sec @ 720p, 6-DoF camera control",
      "modalities": [
        "video"
      ],
      "capabilities": [
        "multimodal",
        "open_weights"
      ],
      "secondary_tags": [
        "open_weights",
        "multimodal",
        "video",
        "ssm_hybrid"
      ],
      "notable": "Generates 60-second 720p clips in 34 seconds on a single RTX 5090 with NVFP4 — first open world model that fits production budgets per-GPU, 36x throughput over prior open baselines.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "gemini-3-5-flash",
      "name": "Gemini 3.5 Flash",
      "vendor": "Google + DeepMind",
      "release_date": "2026-05-19",
      "branch": "multimodal",
      "parents": [],
      "influences": [],
      "openness": "closed",
      "parameters": "Undisclosed",
      "context": "1M",
      "modalities": [
        "text",
        "image",
        "audio",
        "video"
      ],
      "capabilities": [
        "tools",
        "reasoning",
        "agentic",
        "long_context",
        "multimodal"
      ],
      "secondary_tags": [
        "closed",
        "multimodal",
        "frontier",
        "agentic",
        "ultra_long_context"
      ],
      "notable": "Speed-intelligence Pareto leader at I/O 2026: Terminal-Bench 2.1 76.2%, MCP Atlas 83.6%, GDPval-AA 1656 Elo, AA Intelligence Index 55.3. Pricing $1.50/$9 per Mtok (3x prior Flash list).\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "gemini-omni-flash",
      "name": "Gemini Omni Flash",
      "vendor": "Google + DeepMind",
      "release_date": "2026-05-19",
      "branch": "multimodal",
      "parents": [
        "gemini-3-5-flash"
      ],
      "influences": [],
      "openness": "closed",
      "parameters": "Undisclosed",
      "context": "Undisclosed",
      "modalities": [
        "text",
        "image",
        "audio",
        "video"
      ],
      "capabilities": [
        "multimodal"
      ],
      "secondary_tags": [
        "closed",
        "multimodal",
        "frontier",
        "video"
      ],
      "notable": "Avatar insertion, cinematic editing via prompt; available immediately in Gemini app, Flow, YouTube Shorts and Create; API in 'coming weeks'.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "qwen-3-7-max",
      "name": "Qwen3.7-Max",
      "vendor": "Alibaba",
      "release_date": "2026-05-20",
      "branch": "reasoning",
      "parents": [
        "qwen-3-6-plus"
      ],
      "influences": [],
      "openness": "closed",
      "parameters": "Undisclosed",
      "context": "1M",
      "modalities": [
        "text"
      ],
      "capabilities": [
        "tools",
        "reasoning",
        "agentic",
        "long_context"
      ],
      "secondary_tags": [
        "closed",
        "frontier",
        "agentic",
        "china",
        "ultra_long_context"
      ],
      "notable": "First Chinese model in AA Intelligence Index top 5 (56.6, ahead of Gemini 3.5 Flash at 55.3); demonstrated 35-hour autonomous tool-use run with 1,158 calls.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "cohere-command-a-plus",
      "name": "Command A+",
      "vendor": "Cohere",
      "release_date": "2026-05-20",
      "branch": "mixture_of_experts",
      "parents": [],
      "influences": [],
      "openness": "open_source",
      "parameters": "218B total / 25B active (128 experts, 8 active)",
      "context": "128K",
      "modalities": [
        "text",
        "image"
      ],
      "capabilities": [
        "tools",
        "reasoning",
        "agentic",
        "multimodal",
        "open_weights",
        "long_context"
      ],
      "secondary_tags": [
        "open_source",
        "mixture_of_experts",
        "multilingual",
        "enterprise",
        "rag_native"
      ],
      "notable": "First Apache 2.0 frontier-adjacent MoE from a Western lab. τ²-Bench Telecom jumped 37% to 85%; Terminal-Bench Hard 3% to 25%; AA-Omniscience Non-Hallucination #1 at 86%; 48 languages with citation grounding for sovereign / on-prem RAG.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "fara-1-5-27b",
      "name": "Fara1.5-27B",
      "vendor": "Microsoft",
      "release_date": "2026-05-22",
      "branch": "multimodal",
      "parents": [
        "qwen-3-5"
      ],
      "influences": [],
      "openness": "open_weights",
      "parameters": "27B (dense, Qwen3.5 fine-tune)",
      "context": "Undisclosed",
      "modalities": [
        "text",
        "image"
      ],
      "capabilities": [
        "tools",
        "agentic",
        "multimodal",
        "computer_use",
        "open_weights"
      ],
      "secondary_tags": [
        "open_weights",
        "multimodal",
        "computer_use",
        "us"
      ],
      "notable": "Online-Mind2Web 72% (beats OpenAI Operator 58.3%, Gemini 2.5 Computer Use 57.3%, Yutori Navigator n1 64.7%); WebVoyager 88.6%. Collapses browser-agent fleet cost structure.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "hy-mt2-30b-a3b",
      "name": "Hunyuan Hy-MT2-30B-A3B",
      "vendor": "Tencent",
      "release_date": "2026-05-21",
      "branch": "mixture_of_experts",
      "parents": [],
      "influences": [],
      "openness": "open_source",
      "parameters": "30B total / 3B active",
      "context": "Undisclosed",
      "modalities": [
        "text"
      ],
      "capabilities": [
        "multilingual",
        "open_weights"
      ],
      "secondary_tags": [
        "open_source",
        "mixture_of_experts",
        "multilingual",
        "china",
        "small"
      ],
      "notable": "Outperforms DeepSeek-V4-Pro and Kimi K2.6 in fast-thinking translation mode; FLORES-200 leader among open models; powers new Tencent Hy-Translate mini-program with offline mode.\n",
      "status": "active",
      "placement_confidence": "medium"
    },
    {
      "id": "mellum2",
      "name": "Mellum2",
      "vendor": "JetBrains",
      "release_date": "2026-06-01",
      "branch": "mixture_of_experts",
      "parents": [],
      "influences": [],
      "openness": "open_source",
      "parameters": "12B total / 2.5B active",
      "context": "Undisclosed",
      "modalities": [
        "text"
      ],
      "capabilities": [
        "tools",
        "agentic",
        "open_weights"
      ],
      "secondary_tags": [
        "open_source",
        "mixture_of_experts",
        "code_specialist",
        "agentic",
        "edge"
      ],
      "notable": "Released under Apache 2.0; JetBrains claims competitive benchmark performance with more than 2x faster inference than similar-size models. Important as efficient agent substrate rather than frontier replacement.\n",
      "status": "active",
      "placement_confidence": "medium"
    },
    {
      "id": "cosmos-3-nano",
      "name": "Cosmos 3 Nano",
      "vendor": "NVIDIA",
      "release_date": "2026-06-01",
      "branch": "multimodal",
      "parents": [
        "sana-wm"
      ],
      "influences": [],
      "openness": "open_weights",
      "parameters": "16B (8B reasoner + 8B generator)",
      "context": "Undisclosed",
      "modalities": [
        "text",
        "image",
        "video"
      ],
      "capabilities": [
        "multimodal",
        "reasoning",
        "open_weights"
      ],
      "secondary_tags": [
        "open_weights",
        "physical_ai",
        "world_model",
        "multimodal",
        "synthetic_data"
      ],
      "notable": "Creates an open branch for physical AI reasoning and synthetic-data generation rather than general chatbot competition.\n",
      "status": "active",
      "placement_confidence": "medium"
    },
    {
      "id": "holo-3-1",
      "name": "Holo3.1",
      "vendor": "H Company",
      "release_date": "2026-06-02",
      "branch": "multimodal",
      "parents": [
        "fara-1-5-27b"
      ],
      "influences": [],
      "openness": "open_weights",
      "parameters": "0.8B / 4B / 9B / 35B-A3B",
      "context": "Undisclosed",
      "modalities": [
        "text",
        "image"
      ],
      "capabilities": [
        "tools",
        "agentic",
        "multimodal",
        "computer_use",
        "open_weights"
      ],
      "secondary_tags": [
        "open_weights",
        "computer_use",
        "agentic",
        "edge",
        "quantized"
      ],
      "notable": "Moves computer-use agents toward private/local deployment with multiple model sizes and quantized checkpoints.\n",
      "status": "active",
      "placement_confidence": "medium"
    },
    {
      "id": "claude-fable-5",
      "name": "Claude Fable 5",
      "vendor": "Anthropic",
      "release_date": "2026-06-09",
      "branch": "reasoning",
      "parents": [
        "claude-opus-4-8"
      ],
      "influences": [],
      "openness": "closed",
      "parameters": "Undisclosed (inferred larger than Opus 4.8)",
      "context": "1M tokens",
      "modalities": [
        "text",
        "image"
      ],
      "capabilities": [
        "reasoning",
        "tools",
        "agentic",
        "coding",
        "multimodal"
      ],
      "secondary_tags": [
        "frontier",
        "reasoning",
        "agentic",
        "coding",
        "closed"
      ],
      "notable": "First known frontier model to be force-disabled by a government export-control directive (Jun 12, 2026), three days after GA. Commerce withdrew the order Jun 30, 2026 and Fable 5 was restored globally Jul 1, 2026; Mythos 5 remains limited to approved US organizations.\n",
      "status": "active",
      "placement_confidence": "medium"
    },
    {
      "id": "diffusiongemma",
      "name": "DiffusionGemma",
      "vendor": "Google + DeepMind",
      "release_date": "2026-06-10",
      "branch": "mixture_of_experts",
      "parents": [
        "gemma-4"
      ],
      "influences": [],
      "openness": "open_weights",
      "parameters": "25.2B total / ~3.8B active",
      "context": "256K tokens",
      "modalities": [
        "text",
        "image"
      ],
      "capabilities": [
        "multimodal",
        "open_weights",
        "reasoning"
      ],
      "secondary_tags": [
        "open_weights",
        "diffusion",
        "multimodal",
        "edge"
      ],
      "notable": "First open diffusion LLM with native vLLM support, opening an autoregressive-vs-diffusion architecture split in open weights.\n",
      "status": "active",
      "placement_confidence": "medium"
    },
    {
      "id": "north-mini-code",
      "name": "North Mini Code",
      "vendor": "Cohere",
      "release_date": "2026-06-09",
      "branch": "mixture_of_experts",
      "parents": [
        "command-a"
      ],
      "influences": [],
      "openness": "open_weights",
      "parameters": "~30B total / ~3B active",
      "context": "Undisclosed",
      "modalities": [
        "text"
      ],
      "capabilities": [
        "coding",
        "open_weights",
        "tools"
      ],
      "secondary_tags": [
        "open_weights",
        "coding",
        "edge"
      ],
      "notable": "Cheap self-host coding option in the efficient open-weight tier; not frontier-competitive but a low-cost routing node.\n",
      "status": "active",
      "placement_confidence": "medium"
    },
    {
      "id": "glm-5-2",
      "name": "GLM-5.2",
      "vendor": "Z.AI (Zhipu)",
      "release_date": "2026-06-16",
      "branch": "mixture_of_experts",
      "parents": [
        "glm-5-1"
      ],
      "influences": [],
      "openness": "open_weights",
      "parameters": "~744-753B total / ~40B active",
      "context": "1M tokens",
      "modalities": [
        "text"
      ],
      "capabilities": [
        "reasoning",
        "coding",
        "agentic",
        "open_weights",
        "tools"
      ],
      "secondary_tags": [
        "open_weights",
        "reasoning",
        "coding",
        "agentic",
        "frontier"
      ],
      "notable": "Truly permissive (MIT, no regional limits) frontier-adjacent open model — took the open-weight lead in-window and pressures closed pricing.\n",
      "status": "active",
      "placement_confidence": "medium"
    },
    {
      "id": "minimax-m3",
      "name": "MiniMax-M3",
      "vendor": "MiniMax",
      "release_date": "2026-06-12",
      "branch": "mixture_of_experts",
      "parents": [
        "minimax-m1"
      ],
      "influences": [],
      "openness": "open_weights",
      "parameters": "428B total / ~23B active",
      "context": "1M tokens",
      "modalities": [
        "text",
        "image"
      ],
      "capabilities": [
        "reasoning",
        "multimodal",
        "coding",
        "open_weights"
      ],
      "secondary_tags": [
        "open_weights",
        "reasoning",
        "multimodal",
        "coding"
      ],
      "notable": "Frontier-adjacent coding plus genuine 1M context and native multimodality in one checkpoint, but gated by a non-OSI license.\n",
      "status": "active",
      "placement_confidence": "medium"
    },
    {
      "id": "claude-sonnet-5",
      "name": "Claude Sonnet 5",
      "vendor": "Anthropic",
      "release_date": "2026-06-30",
      "branch": "reasoning",
      "parents": [
        "claude-sonnet-4-6"
      ],
      "influences": [
        "claude-opus-4-8"
      ],
      "openness": "closed",
      "parameters": "Undisclosed",
      "context": "1M tokens",
      "modalities": [
        "text"
      ],
      "capabilities": [
        "reasoning",
        "tools",
        "agentic",
        "long_context"
      ],
      "secondary_tags": [
        "closed",
        "reasoning",
        "agentic",
        "coding",
        "ultra_long_context"
      ],
      "notable": "Default model for Claude Free/Pro at launch; 85.2% SWE-bench Verified and #5 on the AA Intelligence Index (53 at max effort). Independent testing shows it beating Opus 4.8 on agentic knowledge work while costing ~15% more per completed task at max effort — sticker price and cost-per-task formally diverged.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "gpt-5-6-sol",
      "name": "GPT-5.6 Sol",
      "vendor": "OpenAI",
      "release_date": "2026-06-26",
      "branch": "reasoning",
      "parents": [
        "gpt-5-5"
      ],
      "influences": [],
      "openness": "closed",
      "parameters": "Undisclosed",
      "context": "Undisclosed",
      "modalities": [
        "text"
      ],
      "capabilities": [
        "reasoning",
        "agentic",
        "tools"
      ],
      "secondary_tags": [
        "closed",
        "frontier",
        "reasoning",
        "agentic"
      ],
      "notable": "GA 2026-07-09 at Sol $5/$30, Terra $2.50/$15, Luna $1/$6 per M tokens — Terra lands at half GPT-5.5's rate, confirming the closed-lab repricing cycle. Sol leads the AA Coding Agent Index v1.1 at 80 (+2.8 over Claude Fable 5) but trails on SWE-Bench Pro (64.6% vs 80%). METR flagged the highest detected reward-hacking rate it has publicly tested, leaving Sol's autonomy time-horizon estimate spanning 11.3 to 270+ hours. GPT-5.4 retires 2026-07-23.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "longcat-2-0",
      "name": "LongCat-2.0",
      "vendor": "Meituan",
      "release_date": "2026-06-30",
      "branch": "mixture_of_experts",
      "parents": [],
      "influences": [],
      "openness": "open_weights",
      "parameters": "1.6T total / ~33-56B dynamically active",
      "context": "1M tokens",
      "modalities": [
        "text"
      ],
      "capabilities": [
        "agentic",
        "tools",
        "open_weights",
        "long_context"
      ],
      "secondary_tags": [
        "open_weights",
        "agentic",
        "coding",
        "china",
        "ultra_long_context"
      ],
      "notable": "Unmasked as \"Owl Alpha,\" the stealth model that led OpenRouter developer charts for two months. Full weight availability was still settling at release (repo initially said \"coming soon\"; INT8/FP8 uploads observed within days); the domestic-ASIC training claim is vendor-asserted.\n",
      "status": "active",
      "placement_confidence": "medium"
    },
    {
      "id": "leanstral-1-5",
      "name": "Leanstral 1.5",
      "vendor": "Mistral AI",
      "release_date": "2026-06-30",
      "branch": "reasoning",
      "parents": [],
      "influences": [],
      "openness": "open_source",
      "parameters": "119B total / 6.5B active",
      "context": "256K",
      "modalities": [
        "text"
      ],
      "capabilities": [
        "reasoning",
        "agentic",
        "tools",
        "open_weights"
      ],
      "secondary_tags": [
        "open_source",
        "mixture_of_experts",
        "math_specialist",
        "code_specialist",
        "agentic"
      ],
      "notable": "Open state of the art on formal theorem proving — miniF2F saturated (100%) and PutnamBench 587/672 at ~$4/problem vs an estimated $300+ for frontier brute force. Predecessor Leanstral (March 2026) is not in the tree, so parents are left empty.\n",
      "status": "active",
      "placement_confidence": "medium"
    },
    {
      "id": "nemotron-labs-twotower",
      "name": "Nemotron-Labs-TwoTower",
      "vendor": "NVIDIA",
      "release_date": "2026-07-01",
      "branch": "mixture_of_experts",
      "parents": [
        "nemotron-3-nano-omni"
      ],
      "influences": [
        "diffusiongemma"
      ],
      "openness": "open_weights",
      "parameters": "30B total / ~3B active (Nemotron-3-Nano-30B-A3B backbone)",
      "context": "Undisclosed",
      "modalities": [
        "text"
      ],
      "capabilities": [
        "open_weights"
      ],
      "secondary_tags": [
        "open_weights",
        "diffusion",
        "ssm_hybrid",
        "edge"
      ],
      "notable": "Establishes the \"diffusion head on a frozen AR backbone\" retrofit pattern alongside DiffusionGemma — a cheap post-hoc throughput upgrade rather than a rival pretraining paradigm. One checkpoint supports diffusion, mock-AR, and standard AR decoding.\n",
      "status": "active",
      "placement_confidence": "medium"
    },
    {
      "id": "grok-4-5",
      "name": "Grok 4.5",
      "vendor": "xAI",
      "release_date": "2026-07-08",
      "branch": "reasoning",
      "parents": [
        "grok-4-3"
      ],
      "influences": [],
      "openness": "closed",
      "parameters": "Undisclosed",
      "context": "1M",
      "modalities": [
        "text"
      ],
      "capabilities": [
        "reasoning",
        "agentic",
        "tools",
        "coding"
      ],
      "secondary_tags": [
        "closed",
        "agentic",
        "coding",
        "tool_using"
      ],
      "notable": "An \"Opus-class\" workhorse at $2/$6 per M tokens — 4th on the AA Intelligence Index (+16 points over Grok 4.3), 83.3% Terminal-Bench 2.1, but 64.7% SWE-Bench Pro vs Fable 5's 80.4%. Not available in the EU at launch (mid-July target).\n",
      "status": "active",
      "placement_confidence": "medium"
    },
    {
      "id": "hy3",
      "name": "Hy3",
      "vendor": "Tencent",
      "release_date": "2026-07-06",
      "branch": "mixture_of_experts",
      "parents": [
        "hunyuan-2"
      ],
      "influences": [
        "glm-5-2"
      ],
      "openness": "open_weights",
      "parameters": "295B total / 21B active (+3.8B MTP layer)",
      "context": "256K (1M via YaRN in community quants)",
      "modalities": [
        "text"
      ],
      "capabilities": [
        "agentic",
        "tools",
        "open_weights",
        "long_context"
      ],
      "secondary_tags": [
        "open_weights",
        "agentic",
        "china",
        "long_context"
      ],
      "notable": "Leads open models on agentic search (84.2 BrowseComp, 91.0 DeepSearchQA, Tencent-reported) at ~$0.20/$0.80 per M tokens, and set the fastest release-to-local pipeline on record for a 295B-class model: community GGUF quants with 1M context landed ~30 hours after release.\n",
      "status": "active",
      "placement_confidence": "medium"
    },
    {
      "id": "muse-spark-1-1",
      "name": "Muse Spark 1.1",
      "vendor": "Meta",
      "release_date": "2026-07-09",
      "branch": "reasoning",
      "parents": [],
      "influences": [],
      "openness": "closed",
      "parameters": "Undisclosed",
      "context": "Undisclosed",
      "modalities": [
        "text",
        "image"
      ],
      "capabilities": [
        "reasoning",
        "multimodal",
        "agentic",
        "tools"
      ],
      "secondary_tags": [
        "closed",
        "multimodal",
        "agentic"
      ],
      "notable": "The structural shift is distribution, not benchmarks: the first frontier Meta model offered through a first-party API (Meta Model API, public preview) rather than open weights or Meta's own apps.\n",
      "status": "active",
      "placement_confidence": "low"
    },
    {
      "id": "kimi-k3",
      "name": "Kimi K3",
      "vendor": "Moonshot AI",
      "release_date": "2026-07-16",
      "branch": "mixture_of_experts",
      "parents": [
        "kimi-k2-6"
      ],
      "influences": [],
      "openness": "open_weights",
      "parameters": "2.8T total / 104B active (16 of 896 experts)",
      "context": "1M",
      "modalities": [
        "text",
        "image",
        "video"
      ],
      "capabilities": [
        "agentic",
        "tools",
        "multimodal",
        "long_context"
      ],
      "secondary_tags": [
        "open_weights",
        "frontier",
        "china",
        "multimodal",
        "agentic",
        "ultra_long_context"
      ],
      "notable": "Reclassified from closed to open weights on 2026-07-27 when the artifacts and license text landed, and the active-parameter count was corrected from a community estimate of ~50B to a confirmed 104B. The first open-weight-lab model to top an LMArena flagship board (#1 Frontend Code Arena at 1,679 prelim), sitting at 57 on the AA Intelligence Index against roughly 60 for the closed leaders. The license, not the capability, is now the enterprise gate.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "inkling",
      "name": "Inkling",
      "vendor": "Thinking Machines Lab",
      "release_date": "2026-07-15",
      "branch": "mixture_of_experts",
      "parents": [],
      "influences": [],
      "openness": "open_weights",
      "parameters": "975B total / 41B active",
      "context": "1M",
      "modalities": [
        "text",
        "image",
        "audio"
      ],
      "capabilities": [
        "open_weights",
        "multimodal",
        "long_context"
      ],
      "secondary_tags": [
        "open_weights",
        "frontier",
        "us",
        "multimodal",
        "ultra_long_context"
      ],
      "notable": "The largest US-origin open-weights model to date, positioned as a fine-tuning and domain-adaptation base (managed Tinker platform for post-training); an Inkling-Small (276B/12B active) was previewed with weights promised after testing.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "claude-opus-5",
      "name": "Claude Opus 5",
      "vendor": "Anthropic",
      "release_date": "2026-07-24",
      "branch": "reasoning",
      "parents": [
        "claude-opus-4-8"
      ],
      "influences": [
        "claude-fable-5"
      ],
      "openness": "closed",
      "parameters": "Undisclosed",
      "context": "Undisclosed",
      "modalities": [
        "text",
        "image"
      ],
      "capabilities": [
        "reasoning",
        "tools",
        "agentic",
        "multimodal"
      ],
      "secondary_tags": [
        "closed",
        "frontier",
        "adaptive_thinking",
        "agentic",
        "coding",
        "tool_using"
      ],
      "notable": "Near-Fable intelligence at $5/$25 per M tokens — unchanged from Opus 4.8 and roughly half Fable 5 pricing. Anthropic reports state-of-the-art Frontier-Bench and GDPval-AA performance for coding and knowledge work, while Opus 5 remains behind Mythos 5 on cyber capability.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "gemini-3-6-flash",
      "name": "Gemini 3.6 Flash",
      "vendor": "Google + DeepMind",
      "release_date": "2026-07-21",
      "branch": "multimodal",
      "parents": [
        "gemini-3-5-flash"
      ],
      "influences": [],
      "openness": "closed",
      "parameters": "Undisclosed",
      "context": "1M",
      "modalities": [
        "text",
        "image",
        "audio",
        "video"
      ],
      "capabilities": [
        "tools",
        "reasoning",
        "agentic",
        "long_context",
        "multimodal"
      ],
      "secondary_tags": [
        "closed",
        "multimodal",
        "frontier",
        "agentic",
        "efficient",
        "ultra_long_context"
      ],
      "notable": "Launched at $1.50/$7.50 per M tokens, versus 3.5 Flash at $1.50/$9. The economic move is lower completed-task token consumption as well as lower output price; independent fleet-level validation remains pending.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "inkling-small",
      "name": "Inkling-Small",
      "vendor": "Thinking Machines Lab",
      "release_date": "2026-07-30",
      "branch": "mixture_of_experts",
      "parents": [
        "inkling"
      ],
      "influences": [],
      "openness": "open_weights",
      "parameters": "276B total / 12B active (6 of 256 experts plus 2 shared)",
      "context": "1M vendor-stated (Artificial Analysis lists 256K)",
      "modalities": [
        "text",
        "image",
        "audio"
      ],
      "capabilities": [
        "open_weights",
        "multimodal",
        "long_context",
        "tools",
        "agentic"
      ],
      "secondary_tags": [
        "open_weights",
        "us",
        "multimodal",
        "efficient",
        "agentic",
        "coding"
      ],
      "notable": "Fulfills the Inkling-Small weights promised at Inkling's July launch, and lands at 40 on the Artificial Analysis Intelligence Index against Inkling's 41 with under a third of the parameters — the strongest efficiency result among US open-weights models to date. Vendor-claimed SWE-bench Verified 80.2% and GPQA Diamond 89.5%. Context window is recorded with a conflict: the vendor model card says up to 1M while AA's write-up lists 256K, unresolved without AA methodology detail.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "muse-spark-1-2",
      "name": "Muse Spark 1.2",
      "vendor": "Meta",
      "release_date": "2026-08-05",
      "branch": "reasoning",
      "parents": [
        "muse-spark-1-1"
      ],
      "influences": [],
      "openness": "closed",
      "parameters": "Undisclosed",
      "context": "1M",
      "modalities": [
        "text",
        "image"
      ],
      "capabilities": [
        "reasoning",
        "multimodal",
        "agentic",
        "tools",
        "long_context"
      ],
      "secondary_tags": [
        "closed",
        "multimodal",
        "agentic",
        "us"
      ],
      "notable": "The first frontier release this publication has recorded where measured capability and per-task cost rose together at unchanged list pricing. Artificial Analysis, given pre-release access, measured Intelligence Index 54 (xhigh) against 1.1's 51 and a GDPval-AA v2 Elo of 1631, up 260 points, while cost per Index task rose from $0.29 to $0.40 because input tokens rose ~53% and output ~36%. Index v4.1.1 then moved it a further +2.7 points on a grader change alone — the largest single move of any model from that patch, and nearly as much as its own release delivered. Its AA-Omniscience gain (18 to 22) is driven by abstention rather than knowledge: hallucination fell 38% to 28% while the attempt rate fell 82% to 67% and accuracy fell 41% to 38%.\n",
      "status": "active",
      "placement_confidence": "medium"
    },
    {
      "id": "qwen3-8-max",
      "name": "Qwen3.8-Max",
      "vendor": "Alibaba",
      "release_date": "2026-08-03",
      "branch": "mixture_of_experts",
      "parents": [
        "qwen-3-7-max"
      ],
      "influences": [],
      "openness": "closed",
      "parameters": "Vendor-stated 2.4T total / 95B active — uncorroborated",
      "context": "1M",
      "modalities": [
        "text",
        "image",
        "video"
      ],
      "capabilities": [
        "reasoning",
        "multimodal",
        "agentic",
        "tools",
        "long_context"
      ],
      "secondary_tags": [
        "closed",
        "frontier",
        "china",
        "multimodal",
        "agentic",
        "ultra_long_context"
      ],
      "notable": "Independent Intelligence Index of 58 with output speed at 77.6 tok/s, placing it inside the frontier cluster on score. The procurement story is verbosity: Artificial Analysis measured 150M output tokens to run the Index against a class median of 66M, labelled the model \"very verbose\", and spent $1,741.41 evaluating it. Open weights for the Qwen-Max class and a companion Qwen3.8-27B were announced as coming \"next week\" but had not shipped as of 2026-08-08, and AA's FAQ confirms this model is not open source.\n",
      "status": "active",
      "placement_confidence": "medium"
    },
    {
      "id": "lfm2-5-2-6b",
      "name": "LFM2.5-2.6B",
      "vendor": "Liquid AI",
      "release_date": "2026-08-04",
      "branch": "decoder_only",
      "parents": [],
      "influences": [],
      "openness": "open_weights",
      "parameters": "2.6B",
      "context": "128K",
      "modalities": [
        "text"
      ],
      "capabilities": [
        "agentic",
        "tools",
        "efficient"
      ],
      "secondary_tags": [
        "open_weights",
        "us",
        "efficient",
        "agentic",
        "edge",
        "commercial_use_restricted"
      ],
      "notable": "Recorded with a license contradiction rather than a clean open-weight classification. The release page's \"Get Started\" section states the model is \"Open-weight — Download, fine-tune, and deploy without restrictions\", while the LFM Open License v1.0 shipped in the same repository states in Section 5(b) that commercial use by a Legal Entity exceeding the $10M annual revenue Threshold is \"not licensed under this Agreement\". This is stricter than the Kimi K3 terms recorded in W31, which gated commercial hosting above $20M trailing revenue and pointed to a separate commercial agreement; the LFM text names no such path, and Section 11 terminates the license automatically on any non-compliance.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "muse-glimmer-30b",
      "name": "Muse Glimmer",
      "vendor": "Meta",
      "release_date": "2026-08-10",
      "branch": "multimodal",
      "parents": [
        "muse-spark-1-2"
      ],
      "influences": [],
      "openness": "open_weights",
      "parameters": "30B dense (29.6B including vision encoder)",
      "context": "128K",
      "modalities": [
        "text",
        "image"
      ],
      "capabilities": [
        "agentic",
        "tools",
        "multimodal",
        "open_weights",
        "long_context"
      ],
      "secondary_tags": [
        "open_weights",
        "dense",
        "us",
        "edge",
        "agentic",
        "tool_using",
        "multimodal",
        "long_context"
      ],
      "notable": "The license is the release. Every prior Meta open model shipped under a Llama License carrying a 700-million-user gate, a naming rule and an acceptable use policy; Muse Glimmer carries none of them, and Apache 2.0 leaves Meta no lever to withdraw the grant. Artificial Analysis scores it 35 on the Intelligence Index and 44 on the Openness Index, level with DeepSeek V4 Flash (0731) and GLM-5.2. Meta is not serving it on its own API, so every price and latency number a buyer sees comes from a third party rather than from Meta.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "gpt-5-6-cyber",
      "name": "GPT-5.6-Cyber",
      "vendor": "OpenAI",
      "release_date": "2026-08-10",
      "branch": "reasoning",
      "parents": [
        "gpt-5-6-sol"
      ],
      "influences": [],
      "openness": "gated",
      "parameters": "Undisclosed",
      "context": "Undisclosed",
      "modalities": [
        "text"
      ],
      "capabilities": [
        "reasoning",
        "agentic",
        "tools"
      ],
      "secondary_tags": [
        "closed",
        "gated",
        "us",
        "frontier",
        "agentic",
        "tool_using"
      ],
      "notable": "The first frontier model whose distribution mechanism is the product decision. Reported to answer 95.0% of advanced cyber requests against 1.5% for GPT-5.6 Sol under normal safeguards, and rated High rather than Critical for cyber capability under OpenAI's Preparedness Framework. Listed at $12.50 input / $75 output per 1M tokens, by a wide margin the most expensive member of the GPT-5.6 family against Sol at $5 / $30. Placement confidence is medium and openness is recorded as gated because no first-party OpenAI source was located during research and no architecture detail has been disclosed.\n",
      "status": "active",
      "placement_confidence": "medium"
    },
    {
      "id": "grok-4-6",
      "name": "Grok 4.6",
      "vendor": "SpaceXAI",
      "release_date": "2026-08-12",
      "branch": "reasoning",
      "parents": [
        "grok-4-5"
      ],
      "influences": [],
      "openness": "closed",
      "parameters": "Undisclosed",
      "context": "Undisclosed",
      "modalities": [
        "text"
      ],
      "capabilities": [
        "reasoning",
        "agentic",
        "tools",
        "coding"
      ],
      "secondary_tags": [
        "closed",
        "us",
        "frontier",
        "agentic",
        "coding",
        "tool_using",
        "long_horizon"
      ],
      "notable": "Scores 61 on the Artificial Analysis Intelligence Index, tying GPT-5.6 Sol Max and trailing Claude Opus 5 (63) and Fable 5 (62), at $2 input / $6 output against Sol's $5 / $30. The interesting number is not the tie but the turn count: Artificial Analysis measured roughly 53 turns and ~0.5B input tokens to resolve long-horizon agentic tasks against roughly 103 turns and ~2.0B tokens for Claude Opus 5 at max settings. Leads GDPval-AA v2 at 1,753 Elo. Terminal-Bench v3.0 at 26% against 34.6% for GPT-5.6 Sol Max is the gap the release did not close.\n",
      "status": "active",
      "placement_confidence": "medium"
    },
    {
      "id": "gemini-3-7-flash",
      "name": "Gemini 3.7 Flash",
      "vendor": "Google",
      "release_date": "2026-08-13",
      "branch": "multimodal",
      "parents": [
        "gemini-3-6-flash"
      ],
      "influences": [],
      "openness": "closed",
      "parameters": "Undisclosed",
      "context": "1M",
      "modalities": [
        "text",
        "image",
        "audio",
        "video"
      ],
      "capabilities": [
        "reasoning",
        "agentic",
        "tools",
        "multimodal",
        "long_context"
      ],
      "secondary_tags": [
        "closed",
        "us",
        "agentic",
        "coding",
        "tool_using",
        "multimodal",
        "ultra_long_context"
      ],
      "notable": "Priced at $0.75 input / $3.75 output per 1M tokens, half of 3.6 Flash — but Google's own pricing page states this is introductory through 2026-12-31 and that $1.50 / $7.50 applies from 2027-01-01. The doubling is disclosed at launch rather than discovered later, which makes it the cleanest public example of a lab pre-announcing that today's agent unit economics are a promotional rate. Google gave no timeline for the next Pro model, and Pichai declined Pro-cadence questions on the most recent earnings call.\n",
      "status": "active",
      "placement_confidence": "medium"
    },
    {
      "id": "glm-5-3",
      "name": "GLM-5.3",
      "vendor": "Z.AI (Zhipu)",
      "release_date": "2026-08-14",
      "branch": "mixture_of_experts",
      "parents": [
        "glm-5-2"
      ],
      "influences": [],
      "openness": "open_weights",
      "parameters": "753B total (architecture identical to GLM-5.2)",
      "context": "1M",
      "modalities": [
        "text"
      ],
      "capabilities": [
        "reasoning",
        "agentic",
        "tools",
        "coding",
        "long_context"
      ],
      "secondary_tags": [
        "china",
        "frontier",
        "agentic",
        "coding",
        "tool_using",
        "long_horizon",
        "ultra_long_context",
        "open_weights"
      ],
      "notable": "GLM-5.3-Flash MIT weights shipped 2026-08-26 confirming Ox Alpha stealth listing, and the full 753B GLM-5.3 checkpoint followed on Hugging Face at zai-org/GLM-5.3 on 2026-08-27/28, closing the two-week weights promise. The gate is now the license, not availability: GLM-5.3 ships under a bespoke GLM-5.3 license rather than MIT, requiring Z.AI security review before commercial use by any Model-as-a-Service operator with group revenue above $10B over any 12 months. Recorded open_weights because the weights are downloadable; the license restriction is noted here rather than by tagging the row gated.\n",
      "status": "active",
      "placement_confidence": "medium"
    },
    {
      "id": "ornith-1-5-397b",
      "name": "Ornith-1.5",
      "vendor": "Ornith AI",
      "release_date": "2026-08-19",
      "branch": "mixture_of_experts",
      "parents": [],
      "influences": [],
      "openness": "open_weights",
      "parameters": "397B total (35B active); companion 9B distilled variant",
      "context": "256K",
      "modalities": [
        "text"
      ],
      "capabilities": [
        "reasoning",
        "agentic",
        "tools",
        "coding",
        "long_context",
        "open_weights"
      ],
      "secondary_tags": [
        "open_weights",
        "frontier",
        "agentic",
        "coding",
        "tool_using",
        "long_context"
      ],
      "notable": "Recorded at medium placement confidence because every headline score is vendor-reported and no independent reproduction was available at publish. Ornith AI publishes the Terminal-Bench 2.1 result of 89.7% on its own harness; that harness has not been audited by a third party and the tree does not treat the number as comparable to Artificial Analysis or Vals results until it is. New lab under the DeepReinforce umbrella; no prior model on the tree. Reclassify to placement_confidence: high when an independent benchmark result on a public leaderboard exists.\n",
      "status": "active",
      "placement_confidence": "medium"
    },
    {
      "id": "glm-5-3-flash",
      "name": "GLM-5.3-Flash",
      "vendor": "Z.AI (Zhipu)",
      "release_date": "2026-08-26",
      "branch": "multimodal",
      "parents": [
        "glm-5-3"
      ],
      "influences": [],
      "openness": "open_weights",
      "parameters": "320B total / 18B active per token",
      "context": "1M",
      "modalities": [
        "text",
        "image"
      ],
      "capabilities": [
        "reasoning",
        "agentic",
        "tools",
        "coding",
        "multimodal",
        "long_context",
        "open_weights"
      ],
      "secondary_tags": [
        "open_weights",
        "china",
        "frontier",
        "agentic",
        "coding",
        "tool_using",
        "ultra_long_context",
        "mixture_of_experts"
      ],
      "notable": "Confirms the Ox Alpha stealth listing (Aug 20–26) as the same checkpoint. Steady-state economics remain unproven: list API carries a 50% launch promo through Sep 9, 2026, and Artificial Analysis Intelligence Index 57 sits below GLM-5.3 (60) and closed frontier tiers. Weights are datacenter-scale MoE, not edge-deployable; no graded source documents the serving hardware behind the stealth run.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "qwen3-8-flash-next",
      "name": "Qwen3.8-Flash-Next",
      "vendor": "Alibaba",
      "release_date": "2026-08-26",
      "branch": "mixture_of_experts",
      "parents": [
        "qwen3-8-max"
      ],
      "influences": [],
      "openness": "open_weights",
      "parameters": "125B main / 6B active; 51B n-gram embedding params",
      "context": "262K native, 1M via YaRN",
      "modalities": [
        "text",
        "image"
      ],
      "capabilities": [
        "agentic",
        "tools",
        "multimodal",
        "long_context",
        "open_weights"
      ],
      "secondary_tags": [
        "open_weights",
        "china",
        "frontier",
        "agentic",
        "multimodal",
        "ultra_long_context",
        "gated"
      ],
      "notable": "Architecture preview under qwen-community-1.0 license rather than Apache 2.0. Procurement must treat open checkpoint and production API as different SKUs — the hosted API ships 1M context and official tool calling; Alibaba's release post does not publish a cached-input rate.\n",
      "status": "active",
      "placement_confidence": "medium"
    },
    {
      "id": "granite-4-2-30b",
      "name": "IBM Granite 4.2",
      "vendor": "IBM",
      "release_date": "2026-08-25",
      "branch": "reasoning",
      "parents": [
        "granite-4-0"
      ],
      "influences": [],
      "openness": "open_weights",
      "parameters": "3B, 8B, and 30B dense sizes",
      "context": "512K extended",
      "modalities": [
        "text"
      ],
      "capabilities": [
        "reasoning",
        "agentic",
        "tools",
        "coding",
        "long_context",
        "open_weights"
      ],
      "secondary_tags": [
        "open_weights",
        "enterprise",
        "agentic",
        "coding",
        "tool_using",
        "long_context"
      ],
      "notable": "Week's clearest enterprise on-prem agent stack without hyperscaler capex exposure — drowned out by Hot Chips but directly substitutable for hosted coding agents when paired with Granite Speech 5.0 Turbo ASR on the same release day.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "thomson-1-0-small",
      "name": "Thomson-1.0-Small",
      "vendor": "Thomson Reuters",
      "release_date": "2026-08-24",
      "branch": "mixture_of_experts",
      "parents": [
        "qwen-3-6-35b-a3b"
      ],
      "influences": [],
      "openness": "gated",
      "parameters": "35B total / 3B active (Qwen3.6-35B-A3B derivative)",
      "context": "262K native",
      "modalities": [
        "text"
      ],
      "capabilities": [
        "tools",
        "long_context"
      ],
      "secondary_tags": [
        "gated",
        "enterprise",
        "legal",
        "long_context"
      ],
      "notable": "Vertical moat is continual learning on Westlaw and Practical Law corpora, not weight sovereignty — CoCounsel retains multi-model agentic workflows via Anthropic's Claude Agent SDK. Recorded gated because the Hugging Face license is non-commercial academic, not Apache/MIT open weights.\n",
      "status": "active",
      "placement_confidence": "medium"
    },
    {
      "id": "hy4-preview",
      "name": "Tencent Hy4 Preview",
      "vendor": "Tencent",
      "release_date": "2026-08-28",
      "branch": "mixture_of_experts",
      "parents": [],
      "influences": [],
      "openness": "open_weights",
      "parameters": "770B total / 49B active per token",
      "context": "1M",
      "modalities": [
        "text"
      ],
      "capabilities": [
        "agentic",
        "tools",
        "coding",
        "long_context",
        "open_weights"
      ],
      "secondary_tags": [
        "open_weights",
        "china",
        "frontier",
        "agentic",
        "coding",
        "ultra_long_context"
      ],
      "notable": "Internal engineering eval is vendor-run; no independent tracker ranking at publish. 31.8% inference throughput gain attributed to self-optimization tooling — treat headline throughput as vendor-stated until reproduced on a neutral harness.\n",
      "status": "active",
      "placement_confidence": "medium"
    },
    {
      "id": "gpt-6-astra",
      "name": "GPT-6 Astra",
      "vendor": "OpenAI",
      "release_date": "2026-09-03",
      "branch": "reasoning",
      "parents": [
        "gpt-5-6-sol"
      ],
      "influences": [],
      "openness": "closed",
      "parameters": "Undisclosed",
      "context": "1M",
      "modalities": [
        "text",
        "image"
      ],
      "capabilities": [
        "reasoning",
        "agentic",
        "tools",
        "multimodal",
        "long_context",
        "computer_use"
      ],
      "secondary_tags": [
        "closed",
        "frontier",
        "multimodal",
        "extended_thinking",
        "agentic",
        "tool_using",
        "computer_use",
        "ultra_long_context"
      ],
      "notable": "Standard API pricing is $10/$50 per million input/output tokens. The 37.2-point ARC-AGI-3 harness spread is not a model-only comparison; ARC Prize says the closed-ended benchmark is not proof of AGI.\n",
      "status": "active",
      "placement_confidence": "high"
    },
    {
      "id": "claude-fable-5-1",
      "name": "Claude Fable 5.1",
      "vendor": "Anthropic",
      "release_date": "2026-09-01",
      "branch": "reasoning",
      "parents": [
        "claude-fable-5"
      ],
      "influences": [],
      "openness": "closed",
      "parameters": "Undisclosed",
      "context": "Undisclosed",
      "modalities": [
        "text"
      ],
      "capabilities": [
        "reasoning",
        "agentic",
        "tools"
      ],
      "secondary_tags": [
        "closed",
        "frontier",
        "adaptive_thinking",
        "agentic",
        "code_specialist",
        "tool_using"
      ],
      "notable": "GitHub documents 30-day default prompt and output retention for Fable 5 and 5.1 unless an eligible enterprise receives an approved zero-data- retention route; distribution and data policy are part of the SKU.\n",
      "status": "active",
      "placement_confidence": "medium"
    },
    {
      "id": "gemini-3-8-flash",
      "name": "Gemini 3.8 Flash",
      "vendor": "Google",
      "release_date": "2026-09-03",
      "branch": "reasoning",
      "parents": [
        "gemini-3-7-flash"
      ],
      "influences": [],
      "openness": "closed",
      "parameters": "Undisclosed",
      "context": "Undisclosed",
      "modalities": [
        "text",
        "image"
      ],
      "capabilities": [
        "reasoning",
        "agentic",
        "tools",
        "multimodal"
      ],
      "secondary_tags": [
        "closed",
        "frontier",
        "multimodal",
        "adaptive_thinking",
        "agentic",
        "tool_using"
      ],
      "notable": "Introductory provider pricing expires December 31, 2026. Current task economics should not be treated as a defensible 2027 run rate without the post-promotion rate.\n",
      "status": "active",
      "placement_confidence": "medium"
    }
  ]
}