--- title: >- Kimi K3 topped an LMArena flagship board, priced itself onto the closed vendors' rate card — and shipped without weights. Meanwhile the largest US-origin open-weights model actually landed, and OpenAI showed what it keeps for itself. publication: The Model Pulse slug: 2026-W29 issueNumber: 13 isoYear: 2026 isoWeek: 29 cadence: weekly publishedAt: '2026-07-18' periodLabel: Week 29 of 2026 canonicalUrl: https://brianletort.ai/industry/models/2026-W29 pdfUrl: https://brianletort.ai/downloads/model-pulse-2026-W29.pdf schemaVersion: 2026.05.02 treeDelta: added: - kimi-k3 - inkling updated: [] frontierMovements: - modelId: null name: GPT-Red vendor: OpenAI releaseDate: '2026-07-15' tier: specialist architecture: reasoning - modelId: null name: GPT-5.6 family (Sol / Terra / Luna) on Amazon Bedrock vendor: OpenAI via AWS releaseDate: '2026-07-13' tier: frontier architecture: reasoning openWeights: - modelId: kimi-k3 name: Kimi K3 vendor: Moonshot AI releaseDate: '2026-07-16' tier: frontier architecture: moe - modelId: inkling name: Inkling vendor: Thinking Machines Lab releaseDate: '2026-07-15' tier: open_frontier architecture: moe architecturePatterns: - Extreme MoE sparsity converges on ~2-4% activation at trillion scale - 4-bit-native training collapses the release-to-deployment gap - 1M context becomes the default tier — with flat pricing benchmarks: - LMArena Frontend Code Arena - Artificial Analysis Intelligence Index - LMArena Text Arena - Agentic coding suite (vendor-reported, KimiCode harness) scorecardAsOf: '2026-07-18' vendorSignals: - vendor: Moonshot AI date: '2026-07-16' signal: >- K3 pricing resets the Chinese-lab tier: $0.30 cached / $3.00 input / $15.00 output per M tokens — input up 216% and output up 275% versus K2.6, matching Sonnet 5's standard rate card (~50% above its live intro price) - vendor: Moonshot AI date: '2026-07-16' signal: >- K3 launched as 'open' with no weights: hosted-only, license unpublished until the promised Jul 27 drop, HF repo 404'd the day after launch — AA classifies it proprietary in the interim - vendor: DeepSeek date: '2026-07-14' signal: >- Hard retirement of the deepseek-chat / deepseek-reasoner aliases lands Jul 24, alongside the first time-of-day surge pricing from a major API — 2x listed rates during Beijing peak hours — with V4 GA signaled as imminent - vendor: OpenAI date: '2026-07-15' signal: >- GPT-Red stays internal: a model trained at flagship post-training scale that OpenAI states will never be released — its output ships only as hardening folded into released models - vendor: Google date: '2026-07-15' signal: >- Reported in-place refresh of the Gemma 4 weights, FlashAttention-4 kernels, and chat templates across the HF collection — targeting tool-call JSON consistency, vision OCR, and long-prompt prefill (weakly sourced; single secondary blog) --- # Kimi K3 topped an LMArena flagship board, priced itself onto the closed vendors' rate card — and shipped without weights. Meanwhile the largest US-origin open-weights model actually landed, and OpenAI showed what it keeps for itself. *The Model Pulse · Issue 13 · Week 29 of 2026 · Published 2026-07-18* ## Big Read Moonshot's Kimi K3 (Jul 16) is the week's headline and its own cautionary tale. The 2.8T-parameter MoE became the first open-weight-lab model to top an LMArena flagship board (#1 Frontend Code Arena at 1,679 preliminary, ahead of Fable 5 and GPT-5.6 Sol) and debuted #4 on the AA Intelligence Index at 57.1 — but it launched hosted-only, with weights promised by Jul 27 and no license published, so Artificial Analysis classifies it proprietary in the interim. And the price tells the strategic story: $3/$15 per M tokens, flat across the full 1M context, is roughly 3x the K2.6 tier and matches Claude Sonnet 5's standard rate card — though with Sonnet's $2/$10 intro price running through Aug 31, K3 actually sits ~50% above the live street price. The 'end of cheap Chinese AI' read circulated on launch day (the-decoder; Simon Willison called K3 the most expensive Chinese-lab model to date); the procurement translation is ours: budget models on measured cost-per-task, not on lineage — see the AI Stack Weekly's price-band-convergence read for the market-structure claim. The week's actual open-weights event came from Thinking Machines Lab: Inkling (Jul 15), a 975B/41B-active trimodal MoE under clean Apache 2.0 with BF16 and NVFP4 checkpoints and day-0 support in every major serving stack — the largest US-origin open-weights model to date, positioned explicitly as a fine-tuning base rather than a chat flagship. And OpenAI's GPT-Red disclosure (Jul 15) marks the capability-gating precedent to watch: an internal-only red-teaming model trained at flagship post-training scale that beat human red-teamers 84% to 13% on indirect prompt injection and hardened GPT-5.6 Sol to a 0.05% direct-injection failure rate. The operational read across all three: treat 'open' as a license-and-weights fact, not a launch label; make prompt-injection robustness a quantified line item in model selection; and expect the strongest capabilities to increasingly ship as vendor-internal advantages rather than products. ## Tree delta Two rows added: kimi-k3 lands Moonshot's 2.8T MoE on the mixture-of-experts branch, recorded as closed until the promised weights and license actually ship, and inkling adds the largest US-origin open-weights model — a 975B trimodal Apache 2.0 MoE from Thinking Machines Lab. **Added (2):** `kimi-k3`, `inkling`. **Updated:** none. *GPT-Red is excluded from the tree by design — OpenAI states it will never be released, and the tree tracks deployed or deployable models. The reported Gemma 4 in-place refresh is weakly sourced (single secondary blog) and is covered under vendor signals rather than as a tree update.* ## Frontier movements ### GPT-Red **OpenAI** · 2026-07-15 · specialist · reasoning _An internal-only self-play RL red-teaming model, trained at flagship post-training scale, that beat human red-teamers 84% to 13% on indirect prompt injection — and will never be released_ Procurement should now treat prompt-injection robustness as a quantified, vendor-differentiating spec: GPT-Red hardened GPT-5.6 Sol to a 0.05% direct-injection failure rate (6x fewer than the best model four months prior), and its attack corpora have been folded into every OpenAI release since GPT-5.3 — numbers you can demand equivalents of from every vendor. The withholding itself is the second signal: frontier labs are starting to keep their strongest capabilities as internal advantages, so vendor selection increasingly buys the byproducts of models you will never see. All figures are vendor-reported; the technical preprint promised the week of Jul 20 is the verification checkpoint. Source: [OpenAI (vendor-reported; preprint promised week of Jul 20); live Andon Labs vending-agent compromise demo](https://openai.com/index/unlocking-self-improvement-gpt-red). ### GPT-5.6 family (Sol / Terra / Luna) on Amazon Bedrock **OpenAI via AWS** · 2026-07-13 · frontier · reasoning _Full family GA on Bedrock's Responses API endpoint at OpenAI first-party rates that count toward AWS commitments — with 90% prompt-cache discounts via explicit cache breakpoints_ Teams with AWS enterprise commitments can now route frontier OpenAI traffic through existing spend at no rate premium — re-run the make-vs-route math on any direct OpenAI contract renewal. Mind the caveats before migrating: context is capped at 272K on Bedrock, and Sol is limited to us-east-1/us-east-2 (Terra and Luna add us-west-2), so region-sensitive workloads need the smaller tiers. The 90% prompt-cache discount with explicit breakpoints makes cache design, not list price, the dominant cost lever for agent fleets on this endpoint. Source: [AWS What's New; Bedrock model card](https://aws.amazon.com/about-aws/whats-new/2026/07/openai-gpt-sol-terra/). ## Open weights ### Kimi K3 `kimi-k3` **Moonshot AI** · 2026-07-16 · frontier · moe _2.8T MoE (16 of 896 experts, ~50B active), 1M context, Kimi Delta Attention, MXFP4 quantization-aware training — hosted-only at launch, with weights promised by Jul 27_ Hold the 'open' celebration until Jul 27: K3 launched API-only, the license is unpublished, the HF repo 404'd the day after launch, and AA classifies it proprietary in the interim — so treat it as a closed model with an open-weights option pending, and gate any self-hosting plan on the actual license text. The capability is real regardless: #1 on the LMArena Frontend Code Arena (first open-weight-lab model to top a flagship board), #4 on the AA Intelligence Index, and the best overall SWE Marathon score (42.0, vendor-reported on the KimiCode harness). Plan for heavy serving if weights do land — Moonshot recommends 64+ accelerator supernodes — and note it launched with only reasoning_effort=max, so token spend runs verbose (130M tokens on the AA eval vs a 63M peer average). Source: [Moonshot AI; Artificial Analysis; LMArena; Simon Willison](https://www.kimi.com/blog/kimi-k3). ### Inkling `inkling` **Thinking Machines Lab** · 2026-07-15 · open_frontier · moe _The largest US-origin open-weights model: 975B/41B-active trimodal MoE under Apache 2.0, 1M context, 45T training tokens, BF16 + NVFP4 checkpoints, day-0 transformers/vLLM/SGLang/llama.cpp_ If your open-model strategy assumed the frontier-scale open lane belonged to Chinese labs, revise it: Inkling is a clean Apache 2.0, US-origin base explicitly positioned for fine-tuning and domain adaptation, with serving partners (Baseten, Modal, Databricks, TogetherAI, Fireworks) live at launch and a managed post-training platform (Tinker) attached. Budget realistically — ~2TB VRAM for BF16 or ~600GB for the calibrated NVFP4 checkpoint — and watch the previewed Inkling-Small (276B/12B active) as the practical deployment tier for most teams. The day-0 NVFP4 release on GB300-trained infrastructure also signals where open-model serving economics are heading: 4-bit datapaths as the default, not an afterthought. Source: [Thinking Machines Lab via Hugging Face; NVIDIA (GB300 NVL72 training confirmation)](https://huggingface.co/blog/thinkingmachines-inkling). ## Architecture watch ### Extreme MoE sparsity converges on ~2-4% activation at trillion scale _Examples:_ Kimi K3 (16 of 896 experts, ~1.8%), Inkling (41B of 975B, 4.2%), DeepSeek V4 Pro (49B of 1.6T, ~3%). Three independent trillion-scale designs now activate only 2-4% of parameters per token, which means serving cost tracks memory capacity and interconnect bandwidth, not FLOPs — the constraint that matters when sizing inference fleets for these models. It also makes routing stability a first-order engineering concern (Quantile Balancing, Per-Head Muon are the current techniques), because a mis-routed expert at 1.8% activation wastes a much larger share of the useful compute. Capacity planners should price these models on HBM footprint and east-west bandwidth, and treat vendor active-parameter counts as the real sizing number. Source: [Moonshot AI; Thinking Machines Lab; DeepSeek](https://www.kimi.com/blog/kimi-k3). ### 4-bit-native training collapses the release-to-deployment gap _Examples:_ Kimi K3 (MXFP4/MXFP8 QAT from the SFT stage), Inkling (calibrated NVFP4 day-0, GB300-trained). Both of the week's big releases trained with 4-bit precision in the loop rather than quantizing after the fact — K3 ran quantization-aware training from the SFT stage, and Inkling shipped a calibrated NVFP4 checkpoint the same day as BF16. That removes the accuracy-loss lottery that used to sit between an open-weights release and a production deployment, and it couples open models to FP4-capable datapaths. Procurement implication: hardware refresh decisions should now weight native FP4 support heavily, because the open-model ecosystem is standardizing on it at the source. Source: [Moonshot AI; Thinking Machines Lab via Hugging Face](https://huggingface.co/blog/thinkingmachines-inkling). ### 1M context becomes the default tier — with flat pricing _Examples:_ Kimi K3 (KDA, flat $3/$15 across 1M), Inkling (1M), DeepSeek Sparse Attention. Hybrid linear/sparse attention designs (K3's Kimi Delta Attention, DeepSeek Sparse Attention) have made million-token context cheap enough that Moonshot prices it flat — the same $3/$15 per M tokens at 1M as at 1K, with no long-context surcharge. Once long context stops carrying a price penalty, prompt-cache hit rate becomes the dominant cost lever for agent workloads: the difference between a well-designed cache ($0.30 cached-input on K3, 90% discounts on Bedrock GPT-5.6) and a naive one dwarfs the model-choice delta. Architect agent memory around cache breakpoints before optimizing anything else. Source: [Moonshot AI; AWS Bedrock documentation](https://platform.kimi.ai/docs/guide/kimi-k3-quickstart). ## Benchmark moves ### LMArena Frontend Code Arena Kimi K3 debuted at #1 with 1,679 (preliminary) — the first model from an open-weight lab to top an LMArena flagship board, 48 points clear of Claude Fable 5 - Kimi K3: 1,679 (prelim) — new #1 - Claude Fable 5: 1,631 - GPT-5.6 Sol: 1,618 - GLM-5.2: 1,587 Source: [LMArena](https://arena.ai/leaderboard/text). ### Artificial Analysis Intelligence Index K3 entered #4 of 189 at 57.1, above Opus 4.8 — but generated 130M tokens on the eval (63M peer average) at $0.94 per Index task, and AA classifies it proprietary until weights land - Claude Fable 5: 59.9 — retains the lead - GPT-5.6 Sol: 58.9 - Kimi K3: 57.1 — new, #4 of 189 - Claude Opus 4.8: 56 Source: [Artificial Analysis](https://artificialanalysis.ai/models/kimi-k3). ### LMArena Text Arena K3 entered #9 at 1,486±11 — a 28-place jump over K2.6's #37, moving Moonshot from mid-pack to the top-10 conversation tier in one generation - Kimi K3: 1,486±11 — #9, new this week - Kimi K2.6: #37 (prior generation) Source: [LMArena](https://arena.ai/leaderboard/text). ### Agentic coding suite (vendor-reported, KimiCode harness) K3 posted the best overall SWE Marathon score and near-frontier terminal results — but every number comes from Moonshot's own harness, and W28's lesson was that harness choice can swing cost and scores 2x, so compare against Sol's Codex-harness numbers with care - Kimi K3 — SWE Marathon: 42.0 (best overall, vendor-reported) - Kimi K3 — Terminal-Bench 2.1: 88.3 - Kimi K3 — FrontierSWE: 81.2 (Fable 5: 86.6) - Kimi K3 — GPQA-Diamond / BrowseComp: 93.5 / 91.2 Source: [Moonshot AI (vendor-reported, KimiCode harness)](https://www.kimi.com/blog/kimi-k3). ## Tier scorecard _As of 2026-07-18._ | Tier | Leader | Challenger | Note | |---|---|---|---| | Closed frontier | Claude Fable 5 | GPT-5.6 Sol | Fable 5 keeps the AA Intelligence Index lead (59.9 vs 58.9) and the coding crown; Sol's counter this week is distribution (Bedrock GA at first-party rates) and hardening (0.05% direct-injection failure rate off GPT-Red) rather than raw capability. | | Open frontier | GLM-5.2 | Kimi K3 | GLM-5.2 holds on the strength of last week's independent Databricks validation; K3 credibly challenges on scores (#4 AA Index, #1 Frontend Code Arena) but stays challenger until the promised weights and license actually ship Jul 27 — Inkling enters the tier as the cleanest-licensed base at this scale. | | Reasoning | Claude Fable 5 | GPT-5.6 Sol | Unchanged at the top; K3 launched with only reasoning_effort=max and verbose token behavior (130M tokens on the AA eval), so its reasoning economics are unproven until the promised low/high effort settings arrive. | | Coding | Claude Fable 5 | Kimi K3 | K3 displaces GLM-5.2 as challenger: #1 Frontend Code Arena (1,679 prelim vs Fable 5's 1,631) and the best SWE Marathon score, though the agentic-suite numbers are vendor-reported on Moonshot's own harness while Fable 5 holds FrontierSWE (86.6 vs 81.2). | | Multimodal | Gemini 3.5 Flash | Kimi K3 | Gemini 3.5 Pro missed its reported Jul 17 target — the third slip; K3 enters as challenger with text+image+video input at 1M context, and Inkling's trimodal (text/image/audio) Apache 2.0 base gives builders an open multimodal foundation for the first time at this scale. | | Edge / small | Mellum2 | Inkling-Small (previewed) | Inkling-Small (276B/12B active, weights promised after testing) would reset the efficient-deployment tier if it ships as previewed; until then the Hy3 local-quant pipeline from W28 remains the practical near-frontier-at-home path. | ## Vendor signals - **2026-07-16 · Moonshot AI — K3 pricing resets the Chinese-lab tier: $0.30 cached / $3.00 input / $15.00 output per M tokens — input up 216% and output up 275% versus K2.6, matching Sonnet 5's standard rate card (~50% above its live intro price)** - The 'Chinese open models are 10x cheaper' planning assumption is dead: Moonshot is pricing on capability, not on lane. Re-run any routing or budget model built on that assumption — the K2.x tiers remain available as the value option, but the frontier Chinese tier now costs what the US mid-frontier costs. - _Source:_ [Moonshot AI pricing documentation; Simon Willison](https://platform.kimi.ai/docs/guide/kimi-k3-quickstart) - **2026-07-16 · Moonshot AI — K3 launched as 'open' with no weights: hosted-only, license unpublished until the promised Jul 27 drop, HF repo 404'd the day after launch — AA classifies it proprietary in the interim** - 'Open' is becoming a launch-marketing label decoupled from the license-and-weights fact. Contract and compliance teams should gate any K3 self-hosting or fine-tuning commitment on the actual license text, and treat 'weights coming' announcements as vendor roadmap, not product. - _Source:_ [Artificial Analysis; Simon Willison](https://simonwillison.net/2026/Jul/16/kimi-k3/) - **2026-07-14 · DeepSeek — Hard retirement of the deepseek-chat / deepseek-reasoner aliases lands Jul 24, alongside the first time-of-day surge pricing from a major API — 2x listed rates during Beijing peak hours — with V4 GA signaled as imminent** - Surge pricing is a structural first: inference is being priced like a capacity-constrained utility, so cost models that assume flat per-token rates need a time-of-day dimension, and latency-tolerant batch workloads should be scheduled into off-peak windows. The alias retirement is a hard migration deadline — audit for pinned deepseek-chat/deepseek-reasoner IDs now. - _Source:_ [DeepSeek API news](https://api-docs.deepseek.com/news/news260424) - **2026-07-15 · OpenAI — GPT-Red stays internal: a model trained at flagship post-training scale that OpenAI states will never be released — its output ships only as hardening folded into released models** - This is the clearest capability-gating precedent yet: the strongest capabilities may increasingly arrive as vendor-internal advantages you rent the byproducts of, not products you evaluate. Expect competitors to follow, and adjust diligence accordingly — ask vendors what internal-only systems shaped the model you are buying, and demand the quantified safety deltas (0.05% direct-injection failure is now the number to beat). - _Source:_ [OpenAI (vendor-reported)](https://openai.com/index/unlocking-self-improvement-gpt-red) - **2026-07-15 · Google — Reported in-place refresh of the Gemma 4 weights, FlashAttention-4 kernels, and chat templates across the HF collection — targeting tool-call JSON consistency, vision OCR, and long-prompt prefill (weakly sourced; single secondary blog)** - If confirmed, in-place weight refreshes under an unchanged model name break reproducibility assumptions: self-hosted Gemma 4 operators should re-pull stale local caches and pin content hashes, not model names, in any evaluation or compliance pipeline. Treat this as a caution flag until Google documents it first-party. - _Source:_ [ExplainX (secondary; not yet confirmed first-party)](https://explainx.ai/blog/gemma-4-updates-flash-attention-tool-calling-july-2026) ## Watchlist - **By Jul 27 — Kimi K3 weights, license, and tech report.** The week's biggest open question: if clean weights and a permissive license land as promised, K3 becomes the strongest open-weights model on record and the open-frontier scorecard flips; if the date slips or the license restricts, the 'open launch without weights' pattern hardens into a playbook. - **Jul 24 — DeepSeek V4 GA + hard alias retirement.** deepseek-chat and deepseek-reasoner go dark the same window V4 is signaled to arrive ('mid-July' per DeepSeek) — the forced migration will produce the first clean V4 adoption data and test whether surge pricing survives contact with customers. - **Week of Jul 20 — GPT-Red technical preprint.** Every GPT-Red number is currently vendor-reported; the preprint is the first chance for independent scrutiny of the 84%-vs-13% red-teaming claim and the Fake Chain-of-Thought injection class, which together justify making injection robustness a procurement spec. - **By Jul 31 — Gemini 3.5 Pro slip watch.** The leaked Jul 17 target passed — the third slip. Prediction-market odds (~81% by Jul 31 on Polymarket) are the only forward signal available and are speculation-grade; treat any capacity or contract planning around a 2M-context flagship as unanchored until Google commits a date. - **No date announced — Inkling-Small (276B/12B active) weights.** The previewed small tier is the practical deployment target for most teams eyeing the Inkling base — its release (and whether the Apache 2.0 terms carry over) determines whether Thinking Machines' open push reaches beyond 600GB-class serving budgets. ## Changelog - Two tree rows added (kimi-k3, inkling). K3 establishes an openness convention for this publication: models launched hosted-only with weights promised are recorded as closed until the weights and license actually ship — the row will be updated if the Jul 27 drop lands. - The reported Gemma 4 in-place refresh was handled as a vendor signal rather than a tree update due to weak sourcing (single secondary blog, no first-party confirmation). - GPT-Red was covered in frontier movements but excluded from the tree: OpenAI states it will never be released, and the tree tracks deployed or deployable models. --- Source of truth: `src/data/industry/models/2026-W29.ts`. Canonical HTML: . PDF: . Tree: .