Everything, in order

At Dreamforce, Salesforce and NVIDIA announced Koa, Salesforce's first CRM reasoning model for Agentforce, built by post-training NVIDIA Nemotron 3 Super on a proprietary synthetic dataset modeled on ~27 years of CRM deployments (no customer data used). Koa reasons through multi-step enterprise workflows and uses tools to act; on Salesforce's CRM benchmark it matches or exceeds leading models on CRM actions with roughly 3x fewer errors. Salesforce controls the weights and runs inference inside its own trust boundary (weights not released). Available to select pilot customers at launch; general availability expected winter 2026 in U.S. regions.

TypeSafe AI released Jev in early access — the first model in its "System One" line, built for automation workflows rather than chat. Jev maps unstructured state to typed probabilistic decisions and emits parallel structured outputs (JSON / tool calls), trained with Reinforcement Learning for Calibrated Decisions (RLCD), and is positioned as a "frontier-intelligence function call" for agents, classification, and tool use. Proprietary (conditional commercial use); weights not released; context window undisclosed. Priced at $0.042 /1M input with output free on a single TypeSafe AI serverless route at launch. Vendor figures unverified independently.

Agnes AI, a Singapore omni-modal foundation-model lab, surfaced Agnes 3.0 Flash in mid-September 2026. The disclosed open-weights preview checkpoint (Apache 2.0) is a 33B hybrid-attention model with a 262K context and text/image/video input: 54 of 72 decoder layers run a gated delta rule while 18 use standard grouped-query attention, holding down the KV cache at long context. Runs at bf16 on a single H100/H200-class GPU. The production Agnes 3.0 Flash served via the company's API is a separate checkpoint with a 1M-token context. First model tracked from this org. Vendor figures unverified at launch.

From 04:00 UTC on 2026-09-14 DeepSeek routes every `deepseek-v4-pro` request to V4.1-Flash, billed at V4.1-Flash rates, and says this will continue until V4.1-Pro launches. The id still answers, making this a redirect and a deprecation rather than a retirement — DeepSeek reports V4.1-Flash beats V4-Pro on performance, cost, speed, and total time.

Inference.net listed Schematron V2 Turbo and Small, a pair of 3B HTML-to-JSON extraction models in its "workhorse model" line — small purpose-built LLMs sold on cost per unit of work. Both are schema-driven (the extraction target goes in a JSON schema via response_format, not the prompt) with 128K context; Turbo is throughput-optimized at ~4.14 req/s on one H100 and $0.03/$0.15 per Mtok, Small trades throughput for quality on complex schemas at $0.05/$0.23. Proprietary and API-only via Inference.net and OpenRouter. First models tracked from this org.

Moonshot AI released Kimi K2.8 Preview, a mid-tier coding and agentic model positioned between Kimi K3 and Kimi K2.7 Code, with performance Moonshot describes as close to K3 but with significantly more efficient thinking. It brings the K3-series thinking-effort controls (low/high/max, max default), multimodal input (text, image, video) with text output, and a 1M-token context now available across all membership tiers. Served on Kimi Code under model id kimi-for-coding so existing clients pick it up without config changes. Proprietary, preview status; parameters undisclosed; vendor figures unverified at launch.

Sakana AI released Fugu Max v1.0 alongside Fugu Ultra v2 — the same orchestration architecture tuned for cost rather than peak capability, routing across open-weights and specialized models. Priced at $2 input / $6 output per Mtok ($0.25 cached), roughly 40-60% below competing frontier models. Sakana reports best overall score across six benchmarks including Terminal-Bench 2.1, GPQA-Diamond, and AA-LCR. System-level vendor figures; parameters, architecture, and context ceiling undisclosed.

Sakana AI released Fugu Ultra v2.0, the second generation of its frontier-class orchestration model, reporting frontier-level results with no Claude Fable 5, Fable 5.1, or GPT-6 Astra among its agents — routing instead over open-weights and specialized models including NVIDIA Nemotron, which Sakana positions as resilience against single-vendor and export-control risk. Tuned for sustained reasoning over complex visual and structured data (Sakana-reported 48.3 Chartography, 74.3 DeepSWE). $5/$30 per Mtok standard, $10/$45 above 272K context. Figures describe an orchestrated system, not a single set of weights.

Cognition launches SWE-2, a proprietary coding model for long-horizon asynchronous software engineering in Devin. Post-trained from the Kimi K3 2.8T base with additional RL at multi-trillion-parameter scale, it targets the cost-performance Pareto frontier and is available through Devin Desktop and CLI at launch, with rollout to Devin Web and Fusion.

DeepSeek retired the experimental vision variant V4-Flash-Vision-Exp alongside the V4.1-Flash launch, whose native multimodal support absorbs it. The `deepseek-v4-flash-vision-exp` id temporarily routes to V4.1-Flash.

DeepSeek retired V4-Flash alongside the V4.1-Flash launch. For compatibility the `deepseek-v4-flash` API id temporarily routes to V4.1-Flash, so existing integrations keep answering but are served by the newer model.

DeepSeek released DeepSeek-V4.1-Flash under MIT with open weights on Hugging Face — a 552B-backbone MoE activating ~8B parameters on prefill and ~16B on decode, with a 1M-token context and 384K max output, reading text and images. The headline change is memory rather than scale: FP4 quantization plus "pure CSA2" cross-layer attention reuse compress the KV cache to ~890 bytes per token, roughly an 8x reduction over July's V4-Flash, cutting HBM to a quarter for an equivalent conversation state and making million-token agentic runs viable on one node. Off-peak $0.15/$0.60 per Mtok. DeepSeek reports it narrowly edges Claude Opus 5 and GPT-5.6 Sol on DeepSWE — vendor figures, unverified at launch.

Ant Group's inclusionAI released Ling-3.0-flash-VL under MIT — a 124B/5.5B sparse MoE adding native image and video understanding to the Ling-3.0-flash backbone via a ViT encoder, two-layer MLP projector, and VideoRoPE encoding, on the family's 42-layer KDA + Gated MLA hybrid stack. 256K context; reported 42 on the Artificial Analysis Intelligence Index v4.1.1 against 38 for text-only Ling-3.0-flash. Aimed at visual reasoning and GUI-agent interaction; free on OpenRouter at launch.

Nex AGI releases Nex-N2.5-Max, a trillion-scale text-only MoE agentic model: 1.6T total / 49B active with a native 1M-token context and Apache-2.0 weights on Hugging Face. Reasoning and tool use only (no vision); no tracked hosted API route at launch.

Nex AGI releases Nex-N2.5-Pro, the flagship multimodal MoE agentic model of the Nex-N2.5 family: 397B total / 17B active with a 262K context, Apache-2.0 weights on Hugging Face, and stronger computer use, web browsing, and visually grounded agent workflows. Hosted as a free preview on OpenRouter at launch.

Nex AGI releases Nex-N2.5-mini, the lightweight multimodal MoE member of the Nex-N2.5 family: 35B total / 3B active with a 262K context and Apache-2.0 weights, sharing the Pro model's agentic stack at lower inference cost. Free preview on OpenRouter at launch.

OpenBMB released MiniCPM5-2B, a 2.52B-parameter dense on-device model on a standard Llama architecture with a 131K-token context, under Apache 2.0. OpenBMB reports a 53.9 average over a 34-benchmark set (vs 51.1 for Qwen3.5-4B), with 86.5 on AIME 2025 and 2026, 63.8 on HMMT Feb 2026, and 94.6 on MATH-500. Shipped with its training data and a family of deployment builds (base, mid-training, SFT-only, GGUF, MLX, 4-bit GPTQ, and a -DSpark speculative-decoding draft). Vendor figures unverified at launch.

iFLYTEK announces Spark X2.5 293B (293B-A30B), the cloud flagship MoE of the Spark X2.5 family: 256K context, multilingual generation across 200+ languages, and a focus on coding and agent capabilities, trained and served on domestic Chinese compute via the xfyun MaaS platform. MaaS list pricing published in CNY only.

inclusionAI (Ant Group) releases LLaDA2.2-mini, the lightweight member of the LLaDA2.2 diffusion-LLM family: a 16B-total / ~1.4B-active MoE diffusion model with a 128K context and the Levenshtein-Editing agentic stack, Apache-2.0 weights on Hugging Face.

inclusionAI published Ling-3.0-flash-Fin's open weights on Hugging Face (inclusionAI/Ling-3.0-flash-Fin) under the MIT license, delivering the open-weight release announced at the model's Aug 27 launch. Third-party hosting (DeepInfra at $0.06/$0.18 per Mtok) and community GGUF quantizations followed, confirming public availability by 2026-09-04.

Ant Group's inclusionAI releases Ling-3.0-flash-Sante (model id inclusionai/ling-3.0-flash-sante), a health- and medicine-tuned variant of Ling-3.0-flash on the same 124B-total / ~5.1B-active sparse MoE base. Post-trained for medical knowledge reasoning, clinical safety, evidence-based retrieval, and long-horizon medical tasks while retaining general reasoning, coding, and agentic ability. Text-only (no vision), 262,144-token (256K) context, up to 32,768 output tokens, with reasoning and function calling. Available via hosted serverless API (Novita, OpenRouter, Vercel AI Gateway) with a time-limited free launch window (free through Oct 4 on Vercel AI Gateway). Positioned as a developer API, not a medical device; no public benchmark table at launch, and Sante-specific open weights were not confirmed posted (base family is MIT).

OpenAI releases GPT-6 Astra Pro (API id gpt-6-astra-pro), the higher-quality reasoning tier of GPT-6 Astra — the same underlying model served with reasoning.mode set to 'pro' for tougher professional, coding, research, computer-use, and agentic tasks. 1.05M-token context, up to 128K output. Standard pricing $10/$50 per Mtok (cached input $1/Mtok, batch half price), with a Fast mode at ~2x. Access is limited to ChatGPT Pro, Business, and Enterprise users, off by default at launch and enabled per workspace.

MBZUAI's Institute of Foundation Models releases K2-Horizon-375B-A23B, the flagship sparse-MoE checkpoint of the K2 Horizon family: 375B total / 23B active with a native 512K context, Apache-2.0 weights on Hugging Face, and reasoning/tool-call parsers for agentic tool use and long-horizon reasoning.

HUMAIN previews HUMAIN M3, an Arabic-focused frontier MoE model (428B total) built on the MiniMax-M3 lineage with HUMAIN Arabic post-training. Natively multimodal (text plus image/video understanding) with agentic tool use and controllable reasoning modes, offered as a limited/research preview on HUMAIN Node. Proprietary; weights not released.

OpenAI released GPT-6 Astra (API id gpt-6-astra), its new frontier flagship succeeding GPT-5.6 Sol, calling it its most intelligent and aligned model. It targets state-of-the-art computer/browser use, finished professional artifacts, long-session coding in Codex, and defensive cybersecurity. 1M-token context; available via the OpenAI API and Amazon Bedrock (and Azure), plus a GPT-6 Astra Pro tier for ChatGPT Pro/Business/Enterprise (off by default at launch). Standard pricing $10/$50 per Mtok, Fast mode ~$20/$100. Vendor-reported: OSWorld 2.0 72.6%, FrontierMath Tier 4 v2 97.6%, GPQA Diamond 96.0%, Terminal-Bench 4.0 57.7%, ExploitBench 100%; ARC-AGI-3 ~99.9% only under a stateful adapter harness, and it trails Claude Fable 5.1 on HLE with tools (57.2% vs 65.0%). Crosses the Critical cyber threshold, so exploit-creation is gated behind OpenAI's Daybreak program.

Alibaba's Qwen team releases Qwen3.8-Max-0902, a post-training upgrade of the flagship Qwen3.8-Max on the same 2.4T-parameter MoE base (~95B active). It targets engineering-scale coding, long-horizon autonomous development, multi-tool agent orchestration, and native vision (text/image/video in, text out), with a 1M-token context, ~131K max output, an optional 256K chain-of-thought thinking mode, and unchanged $2/$6 per Mtok pricing. Vendor-reported CodeArena rose +22 to 1,691 (first at launch); closed-weight, API-first.

Alongside Gemini 3.8 Flash, Google DeepMind introduces Gemini 3.8 Flash Cyber — a cybersecurity-tuned variant built on the same foundational intelligence but shipping with more permissive cyber mitigations. It is restricted to trusted defenders (government authorities, critical-infrastructure operators, and software maintainers) via the new Fairwind Program. Google reports frontier-level autonomous vulnerability discovery on CyberGym, a >70% success rate on an internal real-world benchmark across 20 programming languages, and 47.2% pass@1 on the external CWE-Bench patching benchmark. Internally, the Chrome Security team reports 2.6x more correct patches than larger commercial models. 1M-token context, multimodal input, text output; not publicly token-billed. Vendor figures unverified by independent replication at launch.

Meta ships Muse Spark 1.3, the successor to Muse Spark 1.2 — a multimodal reasoning model for long-running agentic, multi-agent, and coding workflows that tracks information across extended tasks, reconciles conflicting inputs, and asks for clarification when needed, with an emphasis on concise execution. Text + image input over a 1M-token context, text output. Standard API pricing $1.25 / $4.25 per Mtok in/out ($0.15 cached input), with a lower-cost muse-spark-1.3-contributor data-sharing tier. Served by Meta via OpenRouter.

Google DeepMind ships Gemini 3.8 Flash, the successor to Gemini 3.7 Flash and its fourth Flash model in under four months. It scores 59 on the Artificial Analysis Intelligence Index at high reasoning (up 3 points from 3.7 Flash), 57 at medium and 52 at low, with gains led by agentic evaluations (t^3-Banking tool use +12 points to 45%, Terminal-Bench v2.1 coding, GDPval-AA v2). 1M-token context, multimodal input (text, image, video, speech) with text output. Pricing matches 3.7 Flash at $0.75/$3.75 per Mtok in/out through end of 2026 ($1.50/$7.50 standard), with cached input keeping a 90% discount. Available in the Gemini app (AI Pro/Ultra), AI Mode, and Gemini in Google Sheets, and for developers via Google Antigravity, AI Studio and the Gemini API. Vendor/third-party figures unverified by independent replication at launch.

Anthropic released Claude Fable 5.1 (API id claude-fable-5-1), the GA successor to Fable 5 for agentic coding, long-running problem-solving, and knowledge work. Base pricing is unchanged at $10/$50 per Mtok but cache reads were cut 75% to $0.25/Mtok; 1M-token context, 128K output. Reported results include Terminal-Bench 4.0 55.8% and Humanity's Last Exam 65.0% with tools — vendor figures at launch. Available on the Claude API, AWS, Google Cloud, and Microsoft Azure; first Anthropic release with the EU AI Act text watermark.

Anthropic released Claude Mythos 5.1, the same underlying model as Fable 5.1 with more permissive safeguards, available only through the Cyber Verification Program and Life Sciences Verification Program. Anthropic reports its strongest cyber capabilities to date (Terminal-Bench 4.0 60.9% vs 55.8% for safeguarded Fable 5.1) while remaining in the lower risk tier of its Frontier Compliance Framework. Access limited to vetted US organizations; not publicly token-billed.

Inception released Mercury 2.5 Preview, a diffusion LLM that generates and refines many tokens in parallel (~1,107 tokens/sec on standard GPUs), positioned as the fastest reasoning LLM with quality comparable to cost-optimized frontier models. Tunable reasoning levels, parallel tool calls, schema-aligned JSON; 260K context, 65,536 output tokens. API-only via Inception and OpenRouter; list $0.20/$0.75 per Mtok with a launch promo at $0.04/$0.15.

Z.ai published GLM-5.3's open weights on Hugging Face (zai-org/GLM-5.3) after a roughly two-week safety review it attributed to unexpectedly strong multi-stage exploit-chaining found during evaluation. The weights ship under a bespoke "GLM-5.3 License" — MIT-equivalent for most users, but companies with >$10B revenue over any consecutive 12 months must pass Z.AI's security review before commercial Model-as-a-Service use. GLM-5.3-Flash remains plain MIT.

Tencent Hunyuan released and open-sourced Hy4 preview under Apache-2.0 — a 770B-total / 49B-active MoE flagship with a 1M-token context, aimed at long-horizon software engineering, office/financial analysis, and scientific work. Weights and an FP8 variant on Hugging Face; also served via Tencent Cloud TokenHub and OpenRouter. Preview stage; vendor figures unverified at launch.

Cohere released Parse 5 (parse-v5.0), a 2.3B document-intelligence vision-language model that converts PDFs, slides, and images into structured Markdown with tables, forms, and bounding boxes, across nine languages. Priced at $1.50 per 1,000 pages via API, with Model Vault, SageMaker, and Azure availability. Positioned on price-to-performance.

Ant Group inclusionAI launched Ling-3.0-flash-Fin, a finance-domain-tuned variant of Ling-3.0-flash (124B/5.1B MoE) for investment and banking workflows, with a 256K context and tool calling. Hosted API first with a one-month free OpenRouter window; open weights announced for the week of Aug 31 2026 (not yet posted at launch).

Z.ai released GLM-5.3-Flash, a 320B-total / 18B-active natively multimodal MoE with a 1M-token context and MIT-licensed open weights (zai-org/GLM-5.3-Flash), priced roughly 10x cheaper on input than the text-only GLM-5.3 flagship. Vision is integrated into the coding/agent loop. Vendor benchmarks unverified at launch.

Alibaba announced Qwen3.8-Flash, the productionized, managed Qwen Cloud API twin of the open-weight Qwen3.8-Flash-Next, running the same Qwen4-preview MoE architecture. Defaults to a 1M-token context with built-in tools; list pricing $0.15 input / $0.47 output per Mtok. Recorded proprietary/API-only.

Alibaba released Qwen3.8-Flash-Next, an open-weight 125B/6B sparse MoE previewing the Qwen4 architecture (Gated-DeltaNet + Qwen Sparse Attention, n-gram embedding, multi-token prediction). Native 262K context extensible to 1M via YaRN; weights under the Qwen Community License 1.0. Status: preview.

IBM released Granite 4.2 in 3B, 8B, and 30B sizes under Apache-2.0 — dense decoder-only models with a thinking / non-thinking switch and low-effort reasoning mode, pre-trained on ~15T tokens and post-trained with multi-stage RL for agentic tool use. The 30B reports ~57 on SWE-bench Verified. Open weights on Hugging Face, Ollama, and GitHub.

The smallest Granite 4.2 model (~4B on Hugging Face), aimed at local / edge deployment, with the family's thinking / non-thinking switch and native tool calling. Open weights on Hugging Face, Ollama, and GitHub.

The mid-size member of IBM's Granite 4.2 open reasoning family (~9B on Hugging Face): a dense decoder-only transformer trained to call tools and act inside sandboxed environments for software-engineering, terminal, and search tasks. Open weights on Hugging Face, Ollama, and GitHub.

Thomson Reuters announced Thomson, its first in-house proprietary LLM, trained on decades of Westlaw, Practical Law, Checkpoint, and Reuters content to a "Fiduciary-Grade" standard (~$40M training investment). First deployed inside Tabular Analysis in CoCounsel Legal; size, architecture, and context are undisclosed, and a smaller open-weight variant is being released for academic use. Status: preview.

Apodex, a new startup led by Chen Tianqiao with chief scientist Simon Du, released Apodex 1.1 — a proprietary general agentic flagship built for verifiable long-horizon professional and scientific work via a web workbench and an asynchronous "Agent Team" — alongside the open-weight Apodex 1.1 mini (35B, Apache 2.0, rolling out), the FrontierAgent harness, and an arXiv report. Parameter count for the flagship is undisclosed.

The 35B open-weight member of the Apodex 1.1 release, designed to run locally under the open-source FrontierAgent harness as a ReAct or multi-agent Agent Team. Announced under Apache 2.0 with weights and developer docs still rolling out at launch.

DeepSeek released DeepSeek-V4-Flash-Vision-Exp, an experimental multimodal (text + image -> text) checkpoint live on the DeepSeek API. It extends the 284B-total / 13B-active V4-Flash MoE with image understanding (up to 600 images per request) over a 1M-token context, matching V4-Flash on text while making a major leap on multimodal agent benchmarks, priced at $0.22/$0.66 per Mtok. Weights were not published at launch.

Tencent Hunyuan open-weighted Hy-MT2-30B-A3B, a fast-thinking multilingual machine-translation MoE (30B total / ~3B active) covering 33 language pairs plus Chinese-dialect and minority-language pairs, with an 8K context small enough to run locally.

The smallest Ornith-1.5 model, a 9B dense coding/agent model, released under MIT with weights on Hugging Face and a quantized Ornith-1.5-9B-Mobile build for iPhone and Android. Vendor-reported: 47.0 Terminal-Bench 2.1 and 70.6 SWE-Bench Verified.

DeepReinforce released Ornith-1.5 in three MIT-licensed scales — a 397B MoE flagship, a 35B MoE (3B active), and a 9B dense model with a mobile build — with weights on Hugging Face. The family extends Ornith-1.0's self-scaffolding into a closed self-improvement loop: the model proposes its own progressively harder tasks, generates an orchestration scaffold for each, and produces the RL rollouts. Vendor-reported (five-run-avg): the 397B scores 85.1 Terminal-Bench 2.1 and 56.0 DeepSWE, which Ornith puts on par with Claude Opus 4.8 (85.0 / 59.0).

The mid-size Ornith-1.5 model, a 35B MoE activating ~3B parameters per token, released under MIT with weights on Hugging Face. Vendor-reported: 68.5 Terminal-Bench 2.1 and 79.0 SWE-Bench Verified, which DeepReinforce reports as beating dense models of similar or larger size.

Z.ai's GLM-5.2 Turbo (API id glm-5.2-fast) surfaced as a speed-optimized hosted serving tier of GLM-5.2, carrying the 1M-token context and served through Z.ai and SCX.ai at premium fast-tier pricing (~$1.99 in / $6.16 out per Mtok). Parameter count, architecture detail, and any open-weight release for the Turbo tier are undisclosed.

Alibaba published the open weights of its Qwen3.8-Max flagship on Hugging Face and ModelScope as Qwen3.8-2.4T-A95B under Apache 2.0 — the same 2.4T-total / ~95B-active MoE, now downloadable with a 262K context extendable to 1M via YaRN, alongside the smaller Qwen3.8-27B.

Dots Studio (rednote-hilab), the AI lab of Xiaohongshu (RedNote), released Dots3-Note Preview, a ~280B / ~16B-active MoE with a 512K context and text+vision+audio input, under Apache-2.0. Introduces TEMPO, an RL method for long-horizon agents that checkpoint their own progress.

Alibaba published Qwen3.8-27B on Hugging Face under Apache 2.0 — the single-GPU open sibling of Qwen3.8-Max. A 27B dense, hybrid-attention model with a native 262K context (extendable to 1M via YaRN) and multimodal (text/image/video) input, promoting the earlier announced-only entry to generally available.

Z.ai released GLM-5.3, a 743B coding model reusing the GLM-5.2 base with expanded post-training, keeping the 1M-token context and 128K output. Launched via the GLM Coding Plan and ZCode; open weights held back pending a safety review (~2 weeks out).

DeepSeek pinned the dated GA build behind its deepseek-v4-pro API id (OpenRouter listing dated Aug 12 2026), the Pro-tier counterpart to V4-Flash-0731. Keeps the V4-Pro 1.6T/49B MoE architecture and 1M-token context; cache-heavy list pricing. Recorded proprietary/API-only (no confirmed separate 0813 open-weight repo).

Google released Gemini 3.7 Flash, a fast, low-cost workhorse tuned for coding and agents: 1M-token context, 65K output, multimodal input, and $0.75/$3.75 per Mtok introductory pricing through 2026-12-31. Reported DeepSWE v1.1 jumped to 65.3% from 49.0% on Gemini 3.6 Flash.

Liquid AI released LFM2.5-VL-3B, a 3.1B open-weight edge vision-language model on the LFM2.5-2.6B base with a SigLIP2 400M vision encoder, tuned for on-device screen understanding, grounding, and tool calling. Reported 80.7 on ScreenSpot-v2 and 69.4 averaged across 28 benchmarks; runs at ~228 tok/s on an Apple M5 Max.

Cohere released North Micro Vision Instruct, a 2.4B Apache-2.0 document-focused vision-language model with a native-resolution vision encoder and a Command A+ language backbone. Strong on document tasks (0.921 DocVQA, 0.808 ChartQA) and grounding (0.732 RefCOCO), with open weights on Hugging Face.

xAI released Grok 4.6, a post-training update on the reported 1.5T-parameter V9 foundation behind Grok 4.5, with a 500K context, configurable reasoning effort, and $2/$6 per Mtok list pricing below 200K tokens.

Microsoft released MAI-Code-1.1-Flash, an efficiency-focused agentic coding model rolling out across GitHub Copilot (CLI and VS Code). Microsoft reports +22% on Terminal-Bench 2.1 and +15% on .NET over MAI-Code-1.0, with 25% greater token efficiency at roughly a quarter of the cost.

Sakana AI released Namazu, a Japanese-specialized reasoning/agent model built on Moonshot's open-weight Kimi K2.6, over a 262K context with native web-search and code-execution tools, priced at $0.95/$4.00 per Mtok. Not available in the EU/EEA, UK, or Switzerland.

NVIDIA released Nemotron 3.5 Lightning, a hybrid Mamba-2 + MoE + Attention model (~31.6B total / ~3.6B active) with a 1M-token context, under OpenMDW-1.1 with weights, data, and recipes — shipped alongside the NeMo Switchyard router.

OpenAI launched GPT-5.6-Cyber, a cybersecurity-specialized model built on GPT-5.6-Sol with a lower refusal rate on dual-use security tasks, reporting a 95% completion rate on advanced exploit-chain prompts. Access is gated to vetted defenders through the new Daybreak Red tier; the model is not generally available and weights are not released.

Meta released Muse Glimmer, a ~30B dense agentic model with a ~1.8B vision encoder, under Apache 2.0 — its first open-weight model after the closed Muse Spark line. Multimodal (text + image input, text output) across 100+ languages, 131K context, and tuned to run on a single consumer GPU or a Mac.

Upstage released Solar Pro 4, an agent-first reasoning workhorse with a 524K-token context and up to 131K output tokens, priced at $0.30/$1.20 per Mtok (launch promo $0.03/$0.12). Upstage is the first Korean company registered as an official model provider.

Ant Group inclusionAI open-weighted Ling-3.0-tiny under MIT — a sparse MoE with 7.9B total / ~1.3B active parameters and a 262K context, distinct from the API-only Ling-3.0-flash. Weights ship in BF16, FP8, and INT4.

Liquid AI released LFM2.5-2.6B, a 2.69B dense on-device model on a hybrid conv + GQA stack with a 131K context, built to plan and call tools entirely on-device. Open weights under the LFM Open License, shipped day-one in GGUF, MLX, and ONNX.

Meta shipped Muse Spark 1.2, a flagship coding model with 1M-token context and multimodal input, priced at $1.25/$4.25 per Mtok (plus a 12x-cheaper contributor tier), alongside Muse Code, a terminal-based coding agent powered by the model.

Qwen3.8-Max went GA on 2026-08-03, two weeks after its WAIC preview: a 2.4T-parameter MoE (~95B active) with 1M-token context, multimodal input, and firm pricing of $2/$6 per Mtok. Alibaba committed to open weights for Qwen3.8-Max and a smaller Qwen3.8-27B checkpoint.

Alongside Qwen3.8-Max GA, Alibaba announced a smaller open-weight Qwen3.8-27B checkpoint with a firm open-weights timeline — its first for the 3.8 generation.

The deepseek-v4-flash API endpoint is silently upgraded to the retrained DeepSeek-V4-Flash-0731 production build. Same endpoint, key, and model name; pricing unchanged at $0.14/$0.28 per Mtok.

DeepSeek ships DeepSeek-V4-Flash-0731, the production build of V4-Flash: the April preview retrained on an improved post-training pipeline for coding, agents, reasoning, and tool use, with the architecture untouched (284B total / 13B active MoE, 1M context). DeepSeek reports it beating its own larger V4-Pro-Preview on all nine published agent and coding benchmarks. Weights on Hugging Face under MIT; API pricing held at $0.14/$0.28 per Mtok, and the swap is silent for existing deepseek-v4-flash callers.

Alibaba adds Qwen3.7-Flash, a cost-optimized vision-language reasoning model in the Qwen3.7 line, listed on OpenRouter and API endpoints on July 27 without a flagship announcement, technical report, or benchmark suite. It carries a 1M-token context and up to 65,536 output tokens, and at $0.03 / $0.13 per 1M input/output tokens is the cheapest 1M-context multimodal model available at release — aimed at high-volume multimodal agent workloads (visual coding, screen perception, browser/computer use, search) where cost matters more than peak intelligence. Closed-weights and API-only; architecture is undisclosed, with community speculation pointing to a small sparse-MoE design.

Moonshot AI publishes the full Kimi K3 weights to Hugging Face under a Modified MIT license on July 26 — a day ahead of its announced July 27 target — making the 2.8T-parameter MoE freely downloadable, modifiable, and self-hostable, and cementing K3 as the largest open-weight model publicly available.

Anthropic releases Claude Opus 5, its flagship model for demanding reasoning, autonomous coding, and long-horizon agentic work — pitched as the go-to model for most knowledge work, approaching Fable 5 capability in many categories at about half the price. Adds a five-level 'effort' dial on the Claude API/Platform to trade compute for capability, a 1M-token context at standard pricing, and up to 128K output tokens. Standard pricing $5/$25 per Mtok (matching Opus 4.8), plus a $10/$50 Fast mode; becomes the default for Claude Max subscribers. Anthropic calls it its most aligned Opus model.

Ant Group's inclusionAI lab releases Ling-3.0-flash, a hybrid-reasoning Mixture-of-Experts model with 124B total parameters and ~5.1B active per token (1/64 expert activation), built for production-scale agents. Ant claims it matches or beats its own ~1T-parameter flagship on most benchmarks shown at 1/8 the total and 1/12 the active parameters — a vendor claim with no public benchmark table at launch. Uses a KDA + MLA hybrid-linear attention stack at a reported 5:1 ratio for an economical 256K-token context. Announced as open-weight under Apache 2.0, but weights and a model card were not yet posted to Hugging Face as of July 24; usable only via hosted API, free on OpenRouter and Vercel AI Gateway through August 3 2026.

Poolside releases Laguna S 2.1, a 118B-total / 8B-active open-weight MoE coding model with a 1M-token context, pitched as 'the West's most capable open-weight model' for its weight class. It scores 70.2% on Terminal-Bench 2.1 and tops the published open disclosed-size table on SWE-bench Multilingual at 78.5%. Trained in under nine weeks on 4,096 H200 GPUs, it ships weights on Hugging Face under OpenMDW-1.1 (BF16/FP8/INT4/NVFP4 + GGUF/MLX) and runs at 4-bit on a single NVIDIA DGX Spark. Hosted free at 256K context and paid at full 1M context via OpenRouter ($0.10/$0.20/$0.01 per 1M input/output/cache-read tokens).

A cyber-specialized model built on 3.5 Flash for finding and fixing software vulnerabilities, deployed inside the CodeMender agent and reaching competitive frontier performance on CyberGym. Limited-access pilot restricted to governments and trusted partners.

Google DeepMind ships Gemini 3.6 Flash, a multimodal 1M-context workhorse for agentic workflows that improves on 3.5 Flash in coding, knowledge work, and computer use while cutting output-token usage ~17% at a lower price ($1.50 / $7.50 per Mtok). Available in the Gemini API, Gemini Enterprise, and the Gemini app.

Fastest, most cost-effective 3.5-class model (~350 output tokens/s) for high-throughput agentic workloads, priced at $0.30 / $2.50 per Mtok with a 1M-token context and built-in computer use. Large step up on 3.1 Flash-Lite and beats 3 Flash on several agentic/coding evals.

DeepSeek moves the V4 model family (V4-Pro and V4-Flash) out of preview and into general availability, closing a run of just under three months from the April 24 preview. The legacy deepseek-chat and deepseek-reasoner API aliases are retired on July 24, 2026 with no fallback; production traffic must reference deepseek-v4-pro / deepseek-v4-flash.

V4-Flash, the efficient tier of the V4 family, exits preview alongside V4-Pro as DeepSeek retires its legacy deepseek-chat and deepseek-reasoner aliases (cutoff July 24, 2026).

Alibaba launches Qwen3.8, a 2.4-trillion-parameter fully-multimodal model — more than double the size of its predecessor — which the Qwen team says ranks "second only to Fable 5" on overall performance (internal evals, no independent benchmarks yet). Qwen3.8-Max-Preview is available to developers via Alibaba's Token Plan subscription and the Qoder / QoderWork coding platforms, with open weights promised "soon" but no timeline, license, or architecture details disclosed.

Bloomberg reports Google delayed Gemini 3.5 Pro again after the rebuilt model fell short of internal quality goals on hallucinations and reliability — its third slipped target after June and early July. Google DeepMind's Logan Kilpatrick said on July 21 the company is testing it with partners and hopes to "land soon"; there is still no model card or public API entry.

Largest open-weight model to date: 2.8T-parameter MoE (896 experts, 16 active) with Kimi Delta Attention, native multimodal input, and a 1M-token context. API launched Jul 16 at $3/$15 per Mtok; full weights slated for Jul 27. Reported 93.5% GPQA Diamond (strongest published open-weight result).

Mira Murati's lab ships its first model — a 975B-total / 41B-active multimodal MoE (text/image/audio in, text out) pretrained on ~45T tokens, released under Apache-2.0 with weights on Hugging Face (1M context) and hosted on the Tinker API (256K). Debuts at 41 on the Artificial Analysis Intelligence Index, the leading U.S. open-weights model.

The efficient ~32B-active variant of KAT-Coder V2.5, sharing the 256K context and agentic tool-use focus at roughly a fifth of Pro's price ($0.15/$0.60 per Mtok).

Kuaishou's Kwaipilot team releases KAT-Coder-Pro V2.5, an agentic coding MoE (~72B active) trained with large-scale agentic RL in verifiable repository environments, with a 256K context, 80K max output, and API pricing of $0.74/$2.96 per Mtok via StreamLake, Atlas Cloud, and OpenRouter.

OpenAI moves the GPT-5.6 family to general availability after the June 26 restricted preview. Sol, the flagship for difficult professional, coding, research, computer-use, and tool-heavy work, ships with a 1.05M-token context window, 128K max output, and standard pricing of $5/$30 per Mtok, alongside a new max reasoning effort and ultra subagent mode.

Following U.S. government review and a two-week restricted preview, OpenAI rolls out the full GPT-5.6 family — Sol, Terra, and Luna — to general availability across ChatGPT (Plus, Pro, Business, Enterprise) and the API.

Meta Superintelligence Labs releases Muse Spark 1.1 in US public preview on the Meta Model API, the first time Meta charges for one of its models. A natively multimodal reasoning model (text/image/video/PDF/audio in, text out) with a self-compacting 1M-token context, aimed at agentic and coding workflows and priced at $1.25/$4.25 per Mtok — about a quarter of comparable Anthropic/OpenAI models. Closed weights; benchmarks around the Opus 4.8 / GPT-5.5 tier.

Luna, the fast, most cost-efficient tier of the GPT-5.6 family ($1/$6 per Mtok), moves from restricted preview to general availability alongside Sol and Terra.

Terra, the balanced everyday-work tier of the GPT-5.6 family ($2.50/$15 per Mtok), moves from restricted preview to general availability alongside Sol and Luna, available to Plus, Pro, Business, and Enterprise users in ChatGPT and via the API.

SpaceXAI (the rebranded xAI) releases its most capable model to date — the first Grok trained jointly with Cursor (Anysphere), pitched as "Opus-class" but faster, cheaper ($2/$6 per Mtok), and ~4x more token-efficient than Opus 4.8 on SWE-bench Pro. Available in Grok Build (default), Cursor (all plans), and the SpaceXAI console; initially unavailable in the EU.

CEO Arthur Mensch confirms Mistral is preparing a new open-weight Mixture-of-Experts family — "fat but sparse" — entering early access in July 2026 and aimed at the frontier open-weight tier. No parameter count, benchmarks, license terms, or release date disclosed; tracked as rumored.

A compressed variant of Nemotron-3-Super produced with "Iterative Puzzle", trimming the parent to 75.3B total / 9.3B active while keeping the hybrid Mamba-Transformer LatentMoE design and 1M context. NVIDIA reports ~2x higher server throughput at matched user throughput; released on Hugging Face in BF16/FP8/NVFP4 under OpenMDW-1.1.

Tencent officially launches Hunyuan 3.0 (Hy3), the GA of its rebuilt third-generation model, and open-sources it under Apache-2.0. A 295B-total / 21B-active MoE (plus a 3.8B multi-token-prediction layer) with a 256K context and three selectable fast/slow inference modes; Tencent reports it rivals GLM-5.2 and DeepSeek-V4 and matches or surpasses GPT-5.5 on several science benchmarks, with 78.0 on SWE-bench Verified. Weights on Hugging Face (tencent/Hy3) and ModelScope, with a free OpenRouter route (tencent/hy3:free) through July 21, 2026.

Poolside releases Laguna XS 2.1, an upgraded 33B-A3B open-weight MoE coding model served at 256K context. It raises SWE-bench Multilingual by 5.4 points to 63.1% over XS.2, adds open-weighted DFlash speculator models that roughly double local tokens/sec, and moves to the fully permissive OpenMDW-1.1 license. Weights are on Hugging Face (BF16/FP8/INT4/NVFP4) and it's available free on OpenRouter, with paid API pricing of $0.10/$0.20/$0.05 per 1M input/output/cache-read tokens.

Anthropic says Claude Fable 5 is available globally again on Claude Platform, Claude.ai, Claude Code, and Claude Cowork after export controls were lifted; cloud partner access is being re-enabled and safeguards now route high-risk requests to Opus 4.8.

Anthropic restored Mythos 5 access for approved U.S. organizations and continues expanding the Glasswing trusted-access program, while Mythos remains restricted rather than generally available.

Meituan releases LongCat-2.0, a 1.6T-parameter MoE (~48B active) with a 1M-token context for agentic coding, open-sourced under the MIT license on Hugging Face and GitHub. It was trained and served on a ~50,000-card cluster of domestic Chinese AI chips — the first trillion-parameter model Meituan says completed full-process training and inference on home-grown hardware. Vendor-reported results: 59.5 SWE-bench Pro (vs GPT-5.5's 58.6), 70.8 Terminal-Bench 2.1, and 77.3 SWE-bench Multilingual.

Anthropic releases Claude Sonnet 5, its most agentic Sonnet yet — performance approaching Opus 4.8 at a lower price, made the default model on the Free and Pro plans and available to Max, Team, and Enterprise users. Available in Claude Code and via the Claude API as claude-sonnet-5, with introductory pricing of $2 per Mtok input / $10 per Mtok output through Aug 31, 2026, then standard $3/$15.

Reporting says the White House lifted the ban on Anthropic's models after an agreement on additional safeguards, authorizing Anthropic to return Fable 5 to public release channels while Mythos remains limited to pre-vetted partners.

Base44 (a Wix company) rolls out Base1, a general-purpose 'vibe coding' agent fine-tuned on an open-source foundation model using data from tens of millions of platform interactions, selectable alongside GPT-5.5 and Claude Opus 4.8 in its model picker.

Terra is the mid-tier GPT-5.6 preview variant in the Sol/Terra/Luna rollout.

Luna is the lower-cost GPT-5.6 preview variant, still gated by the government-review access process.

The June 26 government carveout restored only narrow Mythos access; Fable remains withdrawn from broad public availability.

U.S. Commerce officials reportedly allowed Anthropic to restore limited Mythos access under tighter controls, while broader Fable access remains restricted.

Sol is the highest-capability GPT-5.6 preview tier, available only to a small vetted cohort.

The Sol, Terra, and Luna variants entered a tightly restricted preview for vetted customers while U.S. officials review security risks.

Reporting says Google delayed broad Gemini 3.5 Pro release from June while testers continue using it in Antigravity and LMArena.

ByteDance officially releases the Seed 2.1 family, led by Seed 2.1 Pro (Doubao-Seed-2.1-pro): a deep-thinking flagship agent model for the "coding and agent era" with a 256K context and strong image/video understanding. ByteDance positions its coding, agent, and multimodal capabilities as comparable to GPT-5.5, citing the highest score on GDPVal, top-tier Agents' Last Exam results, and SOTA visual/video-understanding benchmarks. Proprietary; served via Doubao and Volcano Engine.

ByteDance releases Seed 2.1 Turbo (Doubao-Seed-2.1-turbo) alongside Seed 2.1 Pro: a low-cost, low-latency tier built for large-scale production, feature-complete and positioned as performance-comparable to Pro, with a 256K context and list pricing roughly half that of the Pro tier. Proprietary; served via Doubao and Volcano Engine.

Jie Tang said a Chinese Fable 5-class model would arrive sooner than Elon Musk's Q1 prediction; no product name or launch date has been confirmed.

Open coding-focused Kimi model with 1T total / 32B active parameters, native image/video input, and always-on thinking mode.

MIT-licensed GLM flagship focused on long-horizon coding, agentic engineering, and IndexShare sparse-attention reuse.

Native multimodal 428B/23B-active model with one-million-token context and MiniMax Sparse Attention.

Access to Fable and Mythos was suspended after government action tied to cybersecurity concerns.

Access suspended three days after launch under an export-control directive citing national security.

First open-weight text-diffusion model at scale, built on the Gemma 4 26B-A4B MoE; denoises text in parallel 256-token blocks for up to ~4x faster generation. Apache-2.0.

Cohere's first developer-focused model and the first in its North code-agent family: a 30B/3B-active MoE for agentic coding with a 256K context. Apache-2.0.

OpenAI previewed GPT-5.6, claiming the largest context window of any frontier model.

Anthropic launched Fable 5 as a safeguarded Mythos-class model for broader use.

Available across the Claude API, AWS, and Microsoft Foundry.

Unisound launches U2, a general-purpose "native agentic" LLM built for execution that can autonomously decompose and complete 100+ step workflows; it reports 87.9 on GPQA Diamond and 75 on SWE-bench Verified with ~25% lower thinking-token use. Available on the Unisound Token Hub.

Largest Nemotron 3 model appears on NVIDIA NIM with downloadable weights, 1M context, and agentic reasoning positioning.

A new GPT-Rosalind update folds in GPT-5.5's agentic coding and tool use, improving medicinal-chemistry and genomics performance while using ~31% fewer tokens; expanded to eligible organizations globally via trusted access.

Dense 12B with a unified, encoder-free multimodal architecture and native audio input; runs on a 16GB laptop. Apache-2.0, 256K context.

Shanghai Innovation Institute's Nex alliance releases Nex-N2-Pro, an open-weight (Apache-2.0) agentic model post-trained on Qwen3.5-397B-A17B (397B total / ~17B active) with an "Agentic Thinking" framework, ~262K context, and text+image input.

Microsoft's first in-house frontier reasoning model: a sparse MoE (~35B active, 256K context) trained without third-party distillation; 97.0% on AIME 2025.

An efficient ~5B-active agentic coding model rolling out in GitHub Copilot; Microsoft reports a +16-point SWE-bench Pro lead over Claude Haiku 4.5.

Lower-cost multimodal sibling of Qwen3.7-Max with text, image, and video input and a 1M-token context.

High-efficiency multimodal sparse-MoE (~196B/11B-active) vision-language model with a 256K context and selectable reasoning tiers, for coding agents and search workflows.

Liquid AI ships its on-device MoE (8.3B total / ~1.5B active) with a 131K context that runs in under ~6GB of memory, under the LFM Open License, scaling pretraining to 38T tokens over the October 2025 LFM2-8B-A1B.

Agentic upgrades and stronger long-running task performance.

Open-weight agentic model focused on software engineering, productivity tasks, and model self-evolution workflows.

Fast, cost-efficient Gemini 3.5 tier with a 1M-token context and text/image/audio/video input; Google says it beats Gemini 3.1 Pro on coding and tool-use while running ~4x faster.

Google said its most powerful model in the works needed until the following month before release.

Proprietary agent-first flagship with a 1M-token context, OpenAI/Anthropic-compatible APIs, and long-horizon tool use.

Endpoint shut down as part of Anthropic's 2026 deprecation calendar.

Sparse-MoE flagship; first Chinese model to reach the global LMArena top tier, reportedly trained at ~6% of comparable frontier pre-training cost.

OpenAI releases GPT-5.5 with an 800K-token input context, 128K-token output limit, and stronger reasoning, coding, and multimodal performance.

Limited TAC preview for specialized authorized cybersecurity workflows, paired with stronger verification and safeguards.

1M-token context at $1.25/$2.50 per million tokens; also live on Microsoft Foundry.

DeepSeek introduced the V4 preview series under MIT, led by V4-Pro at 1.6T total / 49B active parameters.

Tencent releases and open-sources the Hy3 preview, its rebuilt third-generation Hunyuan: a 295B-total / 21B-active MoE with a 256K context, positioned as a leading open reasoning-and-agent model for its size. Vendor-reported scores include 74.4 on SWE-bench Verified, 54.4 on Terminal-Bench 2.0, and 70.2 on WideSearch. Weights on GitHub and Hugging Face under Tencent's community license.

Xiaomi releases its open-weight flagship MiMo-V2.5-Pro: a 1.02T-parameter MoE (~42B active) with hybrid attention and a 1M-token context, tuned for frontier-class agentic coding. MIT-licensed, with weights on Hugging Face.

Alongside the Pro flagship, Xiaomi releases MiMo-V2.5, a ~310B/15B-active sparse-MoE model trained on ~48T tokens with a 1M-token context, under the MIT license.

80B/13B-active Hunyuan MoE model released with open weights and agentic tool-use support.

OpenAI launches GPT-Rosalind, a frontier reasoning model purpose-built for biology, drug discovery, and translational medicine, available as a research preview in ChatGPT, Codex, and the API through a trusted-access program.

The Avocado rumor row is retained for provenance; the public model is tracked separately as Muse Spark.

Meta releases Muse Spark, identified in reporting as the productized model behind Meta AI for U.S. users and the public successor to the Avocado effort.

Frontier model shipped only to ~50 defensive-security partners via Project Glasswing.

Removed from all ChatGPT plans after a Feb 13 deprecation notice.

Gemma 4 introduces advanced reasoning open models in 12B, 26B, and 31B sizes.

Z.ai releases GLM-5V-Turbo, its first natively multimodal vision agent: image/video/text input with agent-oriented output (tool calling, task decomposition, GUI interaction) and a ~203K-token context.

Meta's internal model reportedly missed a March target after underperforming top frontier systems; release timing and branding remain unconfirmed.

Apache-2.0 MoE model focused on reasoning, coding, agentic tasks, and Indian-language performance.

Alibaba expands Qwen3.5 with four small dense models (9B, 4B, 2B, 0.8B) under Apache-2.0. The 9B is rated the most intelligent model under 10B parameters, with native vision and a 262K-token context.

OpenAI updates Codex with GPT-5.3-Codex, improving code quality, repository-scale reasoning, and long-running agentic coding workflows.

Open multimodal K2 upgrade with MoonViT, thinking modes, visual coding, and agent-swarm workflows.

OpenAI releases GPT-5.2-Codex for software engineering agents, with a 400K-token input context and 128K-token output limit.

OpenAI releases GPT-5.2 as a stronger general GPT-5 model for reasoning, coding, vision, instruction following, and long-context analysis.

Open K2 reasoning-agent variant for deep thinking and stable long-horizon tool orchestration.

MIT-licensed hybrid linear-attention checkpoints with a 1M-token context and lower KV-cache use.

Anthropic releases Claude Haiku 4.5 as a faster, lower-cost Claude 4.5 tier for coding, tool-use, and latency-sensitive agents.

Anthropic releases Claude Sonnet 4.5, positioning it as its strongest model for coding, agents, and computer-use workflows at launch.

The 0905 update improves agentic coding and frontend generation while doubling context length.

R2 reportedly switched back to Nvidia for training after Huawei Ascend issues, with domestic hardware still targeted for inference.

The statement is tracked as a rumor until a public Grok 5 model release is confirmed.

Anthropic releases Claude Opus 4.1 with improvements for coding, reasoning, and agentic reliability over Claude Opus 4.

Google makes its more deliberative Gemini 2.5 Deep Think reasoning mode available after previewing it at Google I/O 2025.

Google releases Gemini 2.5 Flash-Lite as the lowest-cost, lowest-latency Gemini 2.5 tier for high-volume production tasks.

Alibaba Qwen releases its large open coding-agent MoE with 480B total / 35B active parameters and a 256K-token native context.

Original open 1T-parameter K2 MoE release optimized for coding, reasoning, and agentic tool use.

Updated efficient multimodal reasoning model with stronger video, high-resolution perception, and lower thinking-token use.

Google brings Gemini 2.5 reasoning improvements to a faster, lower-cost Flash production tier with a 1M-token context.

Open coding LLM trained with repository-level reinforcement learning for issue resolution.

Mistral and All Hands AI release Devstral Small 2505, a 24B Apache-2.0 coding-agent model for repository-level software engineering tasks.

Google releases the mobile-first Gemma 3n E4B variant for efficient on-device multimodal inference.

Mistral releases Medium 3 as a lower-cost enterprise workhorse for coding, STEM, search, and multilingual workloads.

Open audio foundation model for speech recognition, audio QA, captioning, generation, and conversation.

Efficient MIT-licensed vision-language MoE for OCR, video, long documents, and agent tasks.

Meta described Behemoth as a still-training 288B-active / nearly 2T-total teacher model used to distill Llama 4 Scout and Maverick.

Open 16B/3B-active MoE demonstrating Moonshot's scalable Muon optimizer work.

Multimodal reinforcement-learning reasoning model reported to match OpenAI o1 on math, coding, and multimodal reasoning.

Qwen releases the 72B chat-tuned Qwen1.5 checkpoint with 32K context and improved alignment.

01.AI releases the chat-tuned Yi-34B checkpoint alongside quantized chat variants.

The consumer launch that ignited the modern LLM race.

Withdrawn following criticism of confident but inaccurate scientific output.

After a staged rollout that began in Feb 2019 over misuse concerns.