--- title: >- April rewrote the floor: open weights crossed the closed frontier, and the safety-gated frontier became a procurement criterion. publication: The Model Pulse slug: 2026-04-recap issueNumber: 1 isoYear: 2026 isoWeek: 17 cadence: monthly_recap publishedAt: '2026-04-25' periodLabel: April 2026 Recap canonicalUrl: https://brianletort.ai/industry/models/2026-04-recap pdfUrl: https://brianletort.ai/downloads/model-pulse-2026-04-recap.pdf schemaVersion: 2026.05.02 treeDelta: added: - claude-opus-4-6 - qwen-3-6-plus - claude-mythos - glm-5-1 - claude-opus-4-7 - qwen-3-6-35b-a3b - kimi-k2-6 - gpt-5-5 - deepseek-v4-pro - deepseek-v4-flash - deepseek-r2 updated: - claude-opus-4-5 - kimi-k2-5 - gpt-5-4 - qwen-3-5 frontierMovements: - modelId: claude-opus-4-7 name: Claude Opus 4.7 vendor: Anthropic releaseDate: '2026-04-16' tier: frontier architecture: agentic - modelId: gpt-5-5 name: GPT-5.5 vendor: OpenAI releaseDate: '2026-04-23' tier: frontier architecture: reasoning - modelId: claude-mythos name: Claude Mythos Preview vendor: Anthropic releaseDate: '2026-04-07' tier: specialist architecture: agentic - modelId: claude-opus-4-6 name: Claude Opus 4.6 vendor: Anthropic releaseDate: '2026-04-02' tier: frontier architecture: reasoning openWeights: - modelId: deepseek-v4-pro name: DeepSeek-V4 Pro vendor: DeepSeek AI releaseDate: '2026-04-24' tier: open_frontier architecture: moe - modelId: deepseek-v4-flash name: DeepSeek-V4 Flash vendor: DeepSeek AI releaseDate: '2026-04-24' tier: open_frontier architecture: moe - modelId: kimi-k2-6 name: Kimi K2.6 vendor: Moonshot AI releaseDate: '2026-04-20' tier: open_frontier architecture: moe - modelId: glm-5-1 name: GLM-5.1 vendor: Z.AI (Zhipu) releaseDate: '2026-04-08' tier: open_frontier architecture: agentic - modelId: qwen-3-6-35b-a3b name: Qwen3.6-35B-A3B vendor: Alibaba releaseDate: '2026-04-16' tier: open_frontier architecture: moe - modelId: deepseek-r2 name: DeepSeek-R2 vendor: DeepSeek AI releaseDate: 2026-04 tier: reasoning architecture: reasoning architecturePatterns: - Reasoning becomes the default mode, not a separate model. - Million-token context as baseline. - Agentic surface area as a first-class capability. - Capability gating as a market signal. benchmarks: - SWE-Bench Pro (coding) - Long-context retrieval (1M+ tokens, vendor-reported) - Agentic tool-use (vendor + third-party) scorecardAsOf: '2026-04-25' vendorSignals: - vendor: Anthropic date: '2026-04-21' signal: Mythos withheld from public access pending capability-gate review. - vendor: DeepSeek AI date: '2026-04-24' signal: Two-config V4 launch (Pro + Flash) plus same-month R2 release under MIT. - vendor: OpenAI date: 2026-04 signal: >- GPT-4 family deprecation timeline tightened; 5.x consolidated to five tiers (Instant, Thinking, Pro, mini, nano). - vendor: Multiple (open-weights serving) date: 2026-04 signal: >- Inference price cuts of 25-40% across major open-weights serving providers following V4 / K2.6 launches. --- # April rewrote the floor: open weights crossed the closed frontier, and the safety-gated frontier became a procurement criterion. *The Model Pulse · Issue 01 · April 2026 Recap · Published 2026-04-25* ## Big Read April 2026 was the most productive month in frontier model history. Eleven model rows landed in the LLM Evolutionary Tree across eight vendors, including the first open-weights MoE to score frontier-class on SWE-Bench Pro coding (Kimi K2.6 at 58.6 and GLM-5.1 at 58.4, both above GPT-5.4 at 57.7 and Claude Opus 4.6 at 57.3). DeepSeek V4 shipped in two configurations under MIT license with 1M-token context and 1.6T total parameters, moving on-prem coding from 'can we?' to 'which workload first?' Anthropic withheld its Mythos flagship on cyber-capability grounds after UK AISI confirmed autonomous offensive capability — an inflection that turns capability gating from a research-org concern into a procurement diligence requirement. The closed frontier responded: GPT-5.5 (Apr 23) and Claude Opus 4.7 (Apr 16) reset the closed-source ceiling, with adaptive-thinking and ultra-long-context as the new defaults. Read together, April was the month the canopy widened on three axes at once: open-vs-closed parity on coding, reasoning-as-default at every tier, and capability gating as a market signal. ## Tree delta 11 model rows added to the tree in April, spanning frontier closed releases, open-frontier MoEs, gated safety-class previews, and reasoning-tier successors. The decoder-only zone widened most; the multimodal and reasoning branches absorbed every major release. **Added (11):** `claude-opus-4-6`, `qwen-3-6-plus`, `claude-mythos`, `glm-5-1`, `claude-opus-4-7`, `qwen-3-6-35b-a3b`, `kimi-k2-6`, `gpt-5-5`, `deepseek-v4-pro`, `deepseek-v4-flash`, `deepseek-r2`. **Updated (4):** `claude-opus-4-5`, `kimi-k2-5`, `gpt-5-4`, `qwen-3-5`. *The Anthropic line gained two flagship releases plus one gated preview in a single month. DeepSeek shipped V4 in two scales (Pro and Flash) plus a same-month R2 reasoning successor — a cadence no other vendor matched in April.* ## Frontier movements ### Claude Opus 4.7 `claude-opus-4-7` **Anthropic** · 2026-04-16 · frontier · agentic _Adaptive-thinking flagship with ultra-long context and full agentic stack (computer-use + MCP + tool use)._ Reset the closed-frontier reasoning ceiling. The agentic surface area (computer-use, MCP, tool-use) makes Opus 4.7 the default reference for closed-frontier procurement when adaptive thinking and long-horizon agent runs are required. Pair with Sonnet 4.6 for the cost-tiered deployment. Source: anthropic.com release notes, model card. ### GPT-5.5 `gpt-5-5` **OpenAI** · 2026-04-23 · frontier · reasoning _Frontier reasoning flagship with multimodal and ultra-long-context as defaults._ OpenAI's response to the V4 / Opus 4.7 surge. Sets the closed-frontier reasoning bar that the open-frontier challengers (V4 Pro, K2.6, GLM-5.1) measure against on SWE-Bench Pro and GPQA. The 5.x family now spans Instant, Thinking, Pro, mini, and nano tiers — useful for matching workload class to model class. Source: openai.com blog, GPT-5.5 system card. ### Claude Mythos Preview `claude-mythos` **Anthropic** · 2026-04-07 · specialist · agentic _Frontier-class capability withheld from public access on cyber-capability grounds._ The first major model gated by its own developer for safety reasons after independent (UK AISI) red-teaming confirmed autonomous offensive capability. Vendor-risk frameworks now need a capability-gate criterion alongside availability SLAs — diligence on 'what could the next release do that the current one cannot?' is no longer optional. Source: anthropic.com, red.anthropic.com, aisi.gov.uk. ### Claude Opus 4.6 `claude-opus-4-6` **Anthropic** · 2026-04-02 · frontier · reasoning _Predecessor to 4.7; held the SWE-Bench Pro closed-frontier line at 57.3 mid-month._ Useful as the April 02 baseline against which open-weights eventually moved past mid-month. Procurement teams running 4.5 should plan upgrade to 4.7 directly; 4.6 will likely move to maintenance status before mid-May. Source: anthropic.com release notes. ## Open weights ### DeepSeek-V4 Pro `deepseek-v4-pro` **DeepSeek AI** · 2026-04-24 · open_frontier · moe _1.6T MoE, 1M context, MIT license, frontier-class on coding._ The headline release of April. First open-weights model with frontier-class SWE-Bench Pro performance under a permissive license. Procurement teams blocked on closed-source data residency or licensing constraints now have an open frontier alternative for coding workloads. The on-prem floor moved up; the merchant-vs-self-host fork is now real. Source: huggingface.co/deepseek, deepseek.com. ### DeepSeek-V4 Flash `deepseek-v4-flash` **DeepSeek AI** · 2026-04-24 · open_frontier · moe _Cost-tiered V4 sibling for high-throughput inference on smaller fabric._ Open-weights answer to GPT-4o-mini and Claude Sonnet at the inference-cost tier. Useful for batch coding agents and high-volume serving where Pro is overkill. Same MIT license, same 1M context, lower active-parameter cost per token. Source: huggingface.co/deepseek model card. ### Kimi K2.6 `kimi-k2-6` **Moonshot AI** · 2026-04-20 · open_frontier · moe _58.6 on SWE-Bench Pro — the first open-weights model to top the closed leaders on coding._ The benchmark moment. K2.6's coding score sits above GPT-5.4 (57.7) and Claude Opus 4.6 (57.3) under an open-weights license. For coding-heavy procurement reads, K2.6 is the new default open-frontier baseline; closed-source premium needs new justification beyond raw score. Source: moonshot.cn, Artificial Analysis SWE-Bench Pro. ### GLM-5.1 `glm-5-1` **Z.AI (Zhipu)** · 2026-04-08 · open_frontier · agentic _58.4 on SWE-Bench Pro with multi-agent-native architecture and ultra-long context._ Second open-weights model to clear the closed-frontier coding bar this month. The multi-agent-native design point makes GLM-5.1 the open-source reference for agent-of-agents deployments — relevant when coordination overhead matters more than single-shot completion. Source: z.ai, GLM-5.1 model card. ### Qwen3.6-35B-A3B `qwen-3-6-35b-a3b` **Alibaba** · 2026-04-16 · open_frontier · moe _Open-weights MoE successor to Qwen3.5; agentic + multimodal at deployable scale._ The most deployable of April's open releases on commodity inference fabric. 35B total / 3B active makes Qwen3.6-A3B the open multimodal pick when V4 Pro is too large and K2.6 is single-modality. Pairs naturally with Qwen3.6-Plus on the closed-source side for hybrid deployments. Source: qwenlm.github.io, model card. ### DeepSeek-R2 `deepseek-r2` **DeepSeek AI** · 2026-04 · reasoning · reasoning _Open-weights reasoning successor to R1, frontier-class long-context._ Same vendor shipping V4 (general) and R2 (reasoning) inside one calendar month is the cadence story. R2 closes the open-vs-closed reasoning gap that R1 opened in 2025 — open-weights reasoning is now a separate procurement category, not a niche. Source: deepseek.com, huggingface.co/deepseek. ## Architecture watch ### Reasoning becomes the default mode, not a separate model. _Examples:_ Claude Opus 4.7 (adaptive thinking), GPT-5.5, DeepSeek-R2, GLM-5.1. April's flagships ship reasoning behavior as a default rather than a separate o-series-style fork. Adaptive thinking, where the model decides how much to deliberate, is now the closed-frontier expectation; explicit reasoning toggles look dated by month-end. Procurement consequence: the reasoning-vs-non-reasoning split is collapsing — assume reasoning is on, budget tokens accordingly. Source: Vendor model cards (Anthropic, OpenAI, DeepSeek, Z.AI). ### Million-token context as baseline. _Examples:_ DeepSeek-V4 Pro (1M), Kimi K2.6 (1M+), Claude Opus 4.7 (ultra-long), GPT-5.5 (ultra-long). Every flagship released in April clears the 1M-token threshold. Long-context is no longer a differentiator at the frontier; it is the floor. Workloads that previously required RAG architectures may now be expressible as direct in-context loads — re-evaluate the retrieval layer when re-baselining for V4 / K2.6 / Opus 4.7. Source: Model cards; HuggingFace technical reports. ### Agentic surface area as a first-class capability. _Examples:_ Claude Opus 4.7 (computer-use, MCP, tool-use), GLM-5.1 (multi-agent-native), Qwen3.6-Plus (agentic), Kimi K2.6 (agentic-ready). Agentic isn't a benchmark category anymore; it is a model property declared on the card. Computer-use, MCP support, and multi-agent coordination ship with the flagship rather than as a downstream wrapper. Architectural read: the boundary between 'model' and 'agent runtime' is dissolving — procurement should evaluate the full agent surface, not just the LM weights. Source: Vendor model cards; Model Context Protocol announcements. ### Capability gating as a market signal. _Examples:_ Claude Mythos Preview (withheld). Anthropic withholding Mythos after UK AISI red-team findings is the first time a major lab gated a frontier-class release on its own initiative for cyber-capability reasons. The signal: capability evaluation now happens before public release, and 'gated' is a status that procurement teams need to track in vendor risk frameworks alongside 'available' and 'deprecated.' Source: anthropic.com, red.anthropic.com, aisi.gov.uk. ## Benchmark moves ### SWE-Bench Pro (coding) Open weights overtook closed for the first time. Two open-weights models clear both closed leaders. - Kimi K2.6: 58.6 - GLM-5.1: 58.4 - GPT-5.4: 57.7 - Claude Opus 4.6: 57.3 Source: Artificial Analysis SWE-Bench Pro leaderboard, vendor model cards. ### Long-context retrieval (1M+ tokens, vendor-reported) Million-token context lands at the frontier with sub-2% retrieval-error rates across multiple vendors. - DeepSeek-V4 Pro: 1M ctx, 1.4% err - Kimi K2.6: 1M ctx, 1.7% err - Claude Opus 4.7: ultra-long, vendor-reported - GPT-5.5: ultra-long, vendor-reported Source: Vendor technical reports; HuggingFace evaluations. ### Agentic tool-use (vendor + third-party) Closed flagships still lead agentic tool-use; open frontier is one tier behind but closing. - Claude Opus 4.7: Closed leader - GPT-5.5: Closed challenger - GLM-5.1: Open leader (multi-agent-native) - DeepSeek-V4 Pro: Open challenger Source: Vendor agent benchmarks; community evaluations. ## Tier scorecard _As of 2026-04-25._ | Tier | Leader | Challenger | Note | |---|---|---|---| | Closed frontier | Claude Opus 4.7 | GPT-5.5 | Adaptive thinking is the closed-source differentiator this month. | | Open frontier | DeepSeek-V4 Pro | Kimi K2.6 | MIT license + 1M context + frontier coding score is the new open baseline. | | Reasoning | DeepSeek-R2 | GPT-5.5 (thinking mode) | Open-weights reasoning is now its own procurement category. | | Coding | Kimi K2.6 | GLM-5.1 | Open-weights tops both closed leaders on SWE-Bench Pro this month. | | Multimodal | Gemini 3.1 Pro | Qwen3.6-Plus | Closed leader from Q1 still holds; Qwen3.6 narrows the gap on the open side. | | Edge / small | Phi-4-mini | Gemma 4 | Q1 leader carries through April; no major edge-tier release this month. | ## Vendor signals - **2026-04-21 · Anthropic — Mythos withheld from public access pending capability-gate review.** - Sets a precedent for vendor-initiated gating on cyber-capability grounds. Procurement frameworks now need a 'gated' status alongside 'available' / 'deprecated.' Adds a new diligence question: what is the next release the vendor evaluated and chose not to ship? - _Source:_ anthropic.com, red.anthropic.com, aisi.gov.uk - **2026-04-24 · DeepSeek AI — Two-config V4 launch (Pro + Flash) plus same-month R2 release under MIT.** - Cadence and licensing both shifted. A monthly cycle with permissive licensing across coding, reasoning, and inference-tier variants is the new open-frontier benchmark for vendor velocity. Closed-source vendors with quarterly cadence have a velocity gap, not just a capability gap. - _Source:_ deepseek.com release notes; huggingface.co model cards - **2026-04 · OpenAI — GPT-4 family deprecation timeline tightened; 5.x consolidated to five tiers (Instant, Thinking, Pro, mini, nano).** - Pricing migration for GPT-4-era workloads accelerates through Q2. Customers should plan migration to 5.x within 60 days; the tier consolidation simplifies model selection but compresses the price-vs-capability slope on the low end. - _Source:_ openai.com platform changelog - **2026-04 · Multiple (open-weights serving) — Inference price cuts of 25-40% across major open-weights serving providers following V4 / K2.6 launches.** - Open-weights inference economics improved materially this month. Re-baseline cost per token for batch agentic workloads against the new floor; the merchant-vs-self-host break-even moved. - _Source:_ Provider pricing pages; community comparison threads ## Watchlist - **May 1-7 — Closed-frontier response to V4 / K2.6 coding scores..** Anthropic and OpenAI both have credible reasons to push a coding-tuned point release to recover the SWE-Bench Pro lead. Watch for Opus 4.7-Codex or GPT-5.5-Codex within two weeks. - **May 1-14 — Hyperscaler Q1 prints (MSFT / GOOG / META / AMZN)..** AI capex commentary will set the inference-pricing trajectory for the rest of Q2. Tracked in The AI Stack Weekly's capital-flow lens; relevant here as the upstream signal for serving costs. - **May 5-20 — Mythos status update from Anthropic + AISI..** If gating becomes permanent or extended, 'gated' becomes a durable status class. If gating lifts under conditions, a capability-gate playbook emerges — either way, the procurement read changes. - **May 10-25 — Open-source reasoning successor cluster..** DeepSeek-R2 set the bar; Qwen, Z.AI, and Moonshot all have R2-class candidates plausibly within 30 days. Watch for the second open-weights reasoning model to clear o-series parity. - **May 15-31 — First gated-capability standardization proposals..** Mythos creates pressure for a cross-vendor gating taxonomy (capability levels, evaluation protocols, release criteria). Watch for AISI / METR / OpenAI / Anthropic joint statements. ## Changelog - Inaugural issue. Cadence: monthly recap covering April 2026 in one read; weekly cadence begins in May. - Tree delta sourced from content/llm-tree/models.yaml April 2026 release rows. - Benchmark scores cite Artificial Analysis leaderboards and vendor model cards as of publish date. - Scorecard reflects April 2026 close; will refresh weekly as new releases land. --- Source of truth: `src/data/industry/models/2026-04-recap.ts`. Canonical HTML: . PDF: . Tree: .