| 29 Oct 2026 | joey-1 | Base Labs | Language | API | MIT | 16,384 tokens (262k on dedicated deploym | Baseten releases 2B-parameter decision model joey-1 at $0.01 per million tokens |
| 29 Oct 2026 | Mistral Large 4-1T-A52B ↗ | Mistral AI | Multimodal | Open weights | Mistral Ecosystem Open-Weight License | 1M tokens | Mistral releases Mistral Large 4 (Le Chonk) 1-trillion-parameter multimodal model on Hugging Face |
| 29 Oct 2026 | Mistral Large 4 (ML4) | Mistral | Language | Open weights | | | Mistral releases Mistral Large 4 (Le Chonk) 1-trillion-parameter multimodal model on Hugging Face |
| 29 Oct 2026 | Le Chonk | Mistral | Language | Open weights | | | Mistral releases Mistral Large 4 (Le Chonk) 1-trillion-parameter multimodal model on Hugging Face |
| 29 Oct 2026 | Gemini 3.8 Flash | Google | Language | API | | | Zapier adds Gemini 3.8 Flash to workflows via AI by Zapier |
| 28 Oct 2026 | Gemini 4 Carbon | Google | Multimodal | Consumer app | | 1M tokens | Google’s Gemini 4 Carbon ranks #1 on Artificial Analysis Intelligence Index v5 with score of 59 |
| 28 Oct 2026 | NVIDIA-Nemotron-3.5-Transcribe-30B-A4B ↗ | NVIDIA | Multimodal | Open weights | OpenMDW-1.1 | 65,536 tokens | NVIDIA releases open-weight Nemotron 3.5 Transcribe model and Granary v2 dataset |
| 28 Oct 2026 | Humanizer ↗ | Jialinyyzz (unspecified org) | Language | Open weights | | null | Open-source Humanizer model claims 95% human-written rewrite rate, but tests show mixed results |
| 28 Oct 2026 | FlashMap | Polymathic AI, Princeton University, USC ISI, NYU, Flatiron Institute | Language | Open weights | Apache-2.0 | | inclusionAI releases Ling-3.1-flash-1019 with 1M-token window and 560B parameters |
| 28 Oct 2026 | Nemotron 3.5 Transcribe ↗ | NVIDIA | Audio | Open weights | OpenMDW-1.1 | 65,536 tokens | NVIDIA releases open-weight Nemotron 3.5 Transcribe model and Granary v2 dataset |
| 27 Oct 2026 | LFM3 | Liquid AI | Multimodal | Open weights | free commercial use up to $10M annual revenue | 256K tokens | Liquid AI launches LFM3 family of six open-weight 2B-class multimodal models |
| 27 Oct 2026 | LFM3-Omni-2B-GGUF | LiquidAI | Multimodal | Open weights | | | Liquid AI launches LFM3 family of six open-weight 2B-class multimodal models |
| 27 Oct 2026 | Qwen3.8-27B-SLQ ↗ | The Kaitchup | Language | Open weights | Apache 2.0 | | The Kaitchup releases statistically-lossless Qwen3.8-27B GGUF models |
| 27 Oct 2026 | LFM3-1.7B ↗ | Liquid AI | Language | | | 256K tokens | Liquid AI launches LFM3 family of six open-weight 2B-class multimodal models |
| 27 Oct 2026 | Genie 3 | Google | Multimodal | Consumer app | | | Google releases Genie 3 world model with 720p, 24fps limits |
| 27 Oct 2026 | Unlimited-OCR-2 ↗ | Baidu | Multimodal | Open weights | | | Baidu releases Unlimited-OCR-2 with unlimited-horizon parsing via SR-SWA |
| 27 Oct 2026 | Calix Communications Language Model | Calix | Language | Open weights | | | Calix releases communications language model on Hugging Face |
| 27 Oct 2026 | d3-ultra ↗ | vLLM Semantic Router | Multimodal | Open weights | Apache-2.0 | 131,072 tokens | vLLM Semantic Router releases 27B open multimodal decision model d3-ultra |
| 27 Oct 2026 | Genie 4 | Google DeepMind | Multimodal | research preview | | | Google DeepMind releases Genie 4 world model with native audio and Gemini 3.8 Flash guidance |
| 26 Oct 2026 | Fern 0.1 | Mirendil | Code | hosted early access only | | 10M tokens | Mirendil announces Fern v0.1 with 10M-token context window |
| 26 Oct 2026 | GPT-6.1 Terra ↗ | OpenAI | Multimodal | API | | | OpenAI launches GPT-6.1 Terra for faster, better-performing AI agents at half the price |
| 26 Oct 2026 | TimeH2O-v1 ↗ | H2O.ai | Multimodal | Open weights | non-commercial | | H2O.ai’s TimeH2O-v1 tops time-series benchmarks with 91.9% win rate |
| 25 Oct 2026 | Nanbeige4.3-33B | Nanbeige | Language | Open weights | Apache-2.0 | 512K | AtomicChat releases quantized GGUF files for Nanbeige4.3-33B |
| 25 Oct 2026 | Nanbeige4.3-33B | Nanbeige LLM Lab | Language | Open weights | Apache 2.0 | 524,288 tokens | Nanbeige releases Nanbeige4.3-33B as a teacher model for lightweight agents |
| 25 Oct 2026 | MOSS-Image | OpenMOSS | Image | | Apache-2.0 | | OpenMOSS releases MOSS-Image, a 5B DiT model for text rendering |
| 25 Oct 2026 | Nanbeige4.3-33B ↗ | Nanbeige Team | Language | Open weights | | 524288 tokens | Nanbeige releases Nanbeige4.3-33B as a teacher model for lightweight agents |
| 25 Oct 2026 | Kimi K3.1 Pro | Kimi | Language | | | | Moonshot AI releases Kimi K3.1 Pro, a 2.8T open-weight model closing gap with proprietary leaders |
| 25 Oct 2026 | Flora ↗ | Nvidia | Multimodal | Open weights | | | Nvidia open-sources Flora to label robot video 125x cheaper than GPT-6 Astra |
| 25 Oct 2026 | OpenMobile-2 | Nanjing University, SenseTime, and other universities | agents | Open weights | | | OpenMobile-2 open-source smartphone agents score 59.6 on SPA-Bench |
| 25 Oct 2026 | Gemini 3.5 | Google | Multimodal | Consumer app | | | Google unveils Gemini 3.5, Omni video AI and Spark agent at I/O |
| 24 Oct 2026 | Deepseek V4-Flash-1017 | Deepseek | Language | Open weights | | | Tom's Hardware Premium weekly roundup covers TP-Link, Surface Laptop Ultra, tape storage, Deepseek V4-Flash-1017 |
| 24 Oct 2026 | RFD4-Proteina | Nvidia and UW Medicine Institute for Protein Design | other | Open weights | | | Nvidia and University of Washington unveil open-source protein design model RFD4-Proteina |
| 24 Oct 2026 | Claude Haiku 5.5 | Claude | Language | Consumer app | | 1 million tokens | Claude Haiku 5.5: New Features, Pricing and Efficient AI Workflows |
| 23 Oct 2026 | CADRE | Kyutai | Video | Open weights | CC BY-NC 4.0 | | Kyutai releases CADRE open video autoencoders to reduce world model lag |
| 23 Oct 2026 | Nemotron 3.5 Bolt E4B | NVIDIA | Language | Open weights | OpenMDW-1.1 | 64K | NVIDIA releases Nemotron 3.5 Bolt E4B for local voice agents and robots |
| 23 Oct 2026 | Dessa | Block | Audio | | | | Block releases Dessa, a full‑duplex speech model trained only on speech |
| 23 Oct 2026 | Qwen3.8-Flash-Next-Gyro-Coder ↗ | AgentionAI | Code | Open weights | | 65536 | AgentionAI releases Qwen3.8-Flash-Next-Gyro-Coder for 24 GB GPUs |
| 23 Oct 2026 | Deepseek V4.1 Flash (revision 1017) | Deepseek | Multimodal | Open weights | | 1 million tokens | Deepseek launches V4.1 Flash 1017 model with 1 M token context |
| 23 Oct 2026 | Kimi K3.1 Pro ↗ | Moonshot AI | Multimodal | Consumer app | | 1 million tokens | Moonshot AI announces Kimi K3.1 Pro with 1M-token context and 3D model creation |
| 23 Oct 2026 | WeVisOne-VL-4B | Tencent | Multimodal | Open weights | Apache License 2.0 | | Tencent releases WeVisOne-VL 4B vision-language model on Hugging Face |
| 22 Oct 2026 | Canva Craft | Canva Research | Image | Consumer app | | | Canva launches Canva Craft image generation model for small businesses |
| 22 Oct 2026 | Decisions API | OpenAI | other | API | | | OpenAI makes its Decisions API widely available using GPT-6 Luna |
| 22 Oct 2026 | Carbon-X ↗ | Hugging Face Bio | Language | Open weights | Apache 2.0 | 40,960-token | Hugging Face releases Carbon-X, a 14.8B open DNA language model |
| 22 Oct 2026 | Phonon 31m | Gradium | Audio | Open weights | Apache 2.0 | | Gradium open-sources Phonon 31M on-device TTS model under Apache 2.0 |
| 22 Oct 2026 | Cohere Rerank 5 | Cohere AI | other | API | | 32K tokens | Cohere releases Rerank 5 with passage-level context focus and calibrated scores |
| 22 Oct 2026 | Ling-3.0-flash-Sante | inclusionAI | Language | Open weights | | 256K tokens | inclusionAI releases Ling-3.0-flash-Sante, a 124B healthcare model |
| 22 Oct 2026 | OlmoEarth v1.3 | Allen Institute for AI | other | Open weights | OlmoEarth Artifact License | | Ai2 releases OlmoEarth v1.3 with global 10M satellite embeddings |
| 22 Oct 2026 | StormScope EU | NVIDIA | Multimodal | Open weights | | | Nvidia adds StormScope EU model to Earth-2 for European weather forecasts |
| 22 Oct 2026 | Claude Duet | Anthropic | other | Consumer app | | | Anthropic launches Claude Duet, a proactive AI agent app for Pro and Max users |
| 22 Oct 2026 | DiffResearch | arXiv authors (unspecified) | other | Open weights | Apache-2.0 | | Researchers release open DiffResearch framework with diffusion model backbone |
| 22 Oct 2026 | Hy-SimulMT-30B-A3B ↗ | Tencent Hunyuan | Multimodal | Open weights | | | Tencent/Hy-SimulMT-30B-A3B · Hugging Face |
| 21 Oct 2026 | Flint | Recursive Superintelligence | Code | Open weights | Apache 2.0 | 262,144 tokens | Recursive releases Flint, a 40B parameter MoE coding model with 2.5B active parameters per token |
| 21 Oct 2026 | FlintForCausalLM | ggml-org | Language | Open weights | | | Ggml-org adds FlintForCausalLM model to llama.cpp |
| 21 Oct 2026 | Kumo Relational ↗ | NVIDIA | other | Open weights | Apache 2.0 / OpenMDW | 10,000 examples | Nvidia releases Kumo Forecast time-series model with retrieval and interpretability |
| 21 Oct 2026 | Flint | Recursive | Code | Open weights | Apache 2.0 | 512k tokens | Recursive releases Flint, a 40B parameter MoE coding model with 2.5B active parameters per token |
| 21 Oct 2026 | Granite Speech 5.0 2B ↗ | IBM | Audio | Open weights | Apache 2.0 | | IBM releases Granite Speech 5.0 2B for multilingual speech recognition |
| 21 Oct 2026 | Flint 40B-A2.5B ↗ | Recursive | Code | Open weights | Apache 2.0 | 32768 tokens | Recursive releases Flint, a 40B parameter MoE coding model with 2.5B active parameters per token |
| 21 Oct 2026 | Ming-3.0-TTS-0.5B ↗ | inclusionAI | Audio | Open weights | MIT | | InclusionAI releases Ming-3.0-audio-flash, a multilingual audio-language model |
| 21 Oct 2026 | Kumo Forecast ↗ | NVIDIA | other | Open weights | Apache 2.0 | 2048 observations | Nvidia releases Kumo Forecast time-series model with retrieval and interpretability |
| 21 Oct 2026 | Nemotron 3.5 Embed 8B BF16 ↗ | NVIDIA | Multimodal | Open weights | | | NVIDIA expands open models across language, robotics, biology and Quantum Computing |
| 21 Oct 2026 | RFD4-Proteina ↗ | Nvidia | other | Open weights | BSD-3 | | Nvidia and University of Washington unveil open-source protein design model RFD4-Proteina |
| 21 Oct 2026 | Sage-A1.5 | InternScience | Language | Open weights | | 200K tokens | InternScience releases Sage-A1.5 35B agentic model for scientific tasks |
| 21 Oct 2026 | Raon-1 mini preview | KRAFTON AI | Language | Open weights | Apache 2.0 | 262,144 tokens | Krafton releases Raon-1 mini preview, a 30B-A4B MoE model with 262,144 token context |
| 21 Oct 2026 | Voxtral 3 | Mistral AI | Audio | API | | 32k tokens | Mistral AI launches Voxtral 3 voice agent stack |
| 20 Oct 2026 | Eleven Realtime | Elevenlabs | Audio | API | | | Elevenlabs launches two realtime speech-to-speech models for AI conversations |
| 20 Oct 2026 | Spark-X2.6-4B | XHToken | Language | Open weights | Apache 2.0 License | 1,048,576 tokens | XHToken releases Spark-X2.6-4B, a 4B on-device language model |
| 20 Oct 2026 | Cosmos3-H-Dreams | NVIDIA | Video | Open weights | NVIDIA Open Model License | | Nvidia releases Cosmos3-H-Dreams real-time generative surgical simulator |
| 20 Oct 2026 | qev-0.8b ↗ | labqoat | other | Open weights | Apache-2.0 | | Qev-0.8b tops decision model benchmarks and runs on a laptop |
| 20 Oct 2026 | FSR 4.1 | AMD | Image | Consumer app | | | AMD adds FSR 4.1 support for RX 6000 GPUs, Xbox Series X/S and Ryzen AI chips |
| 20 Oct 2026 | Bombadil | Electric Plant | Language | cloud | | | Electric Plant raises $11.2 million seed round to launch Bombadil plant‑signal model |
| 20 Oct 2026 | Ornith-1.5-122B ↗ | Ornith AI | Language | Open weights | MIT | 262144 tokens | Ornith-ai/Ornith-1.5-122B (Qwen finetune) |
| 20 Oct 2026 | Tess-4.5-27B ↗ | Continuity Research | Language | Open weights | Apache 2.0 | 65536 | Continuity Research releases Tess-4.5-27B open-weights coding model |
| 20 Oct 2026 | Drex Max | Nace AI | Multimodal | API | | 262,144 tokens | Nace AI releases Drex Max multimodal decision model for computer-use agents |
| 19 Oct 2026 | Clairic | Cloudflare | other | API | | 65,536-token | Cloudflare releases Clairic, a System One extraction model for faster, cheaper data extraction |
| 19 Oct 2026 | FLUX 3 Max | fal | Video | API | | | Fal launches FLUX 3 Max video model, 22× faster than Seedance 2.5 |
| 19 Oct 2026 | Dream-1 | Dream Labs | other | API | | | Dream Labs raises $90 million seed and launches Dream-1 robot AI API |
| 19 Oct 2026 | Kling AI 4.0 | Kling AI | Video | API | | | Kling AI releases version 4.0 with stereo audio and a lighter Flash variant |
| 19 Oct 2026 | Mercury 3 | Inception | Language | API | | 1 million tokens | Inception launches Mercury 3 diffusion LLM, claims parity with GPT-6 Luna and Claude Haiku 5.5 at 1,000+ tokens/sec |
| 19 Oct 2026 | Pegasus 1.5 | TwelveLabs | Video | API | | | TwelveLabs Pegasus 1.5 becomes available on Amazon Bedrock |
| 19 Oct 2026 | Ming-3.0-audio-tiny ↗ | inclusionAI | Multimodal | Open weights | MIT | | InclusionAI releases Ming-3.0-audio-flash, a multilingual audio-language model |
| 19 Oct 2026 | Ming-3.0-audio-flash ↗ | inclusionAI | Multimodal | Open weights | MIT | | InclusionAI releases Ming-3.0-audio-flash, a multilingual audio-language model |
| 19 Oct 2026 | M3.1 Flash ↗ | MiniMax | Multimodal | API | | one million tokens | MiniMax launches M3.1 Flash model at $0.15 per million input tokens with weights due October 23 |
| 19 Oct 2026 | Muse Wonder and Muse Wonder Full | Meta | other | Consumer app | | | Meta may write a $100 billion-$200 billion check for Muse Wonder, says Zuckerberg |
| 19 Oct 2026 | Maxwell | DoubleAI | Code | | | | Tel Aviv-based doubleAI releases debut code execution model Maxwell and raises $200M-$220M |
| 19 Oct 2026 | DFlash 2 ↗ | Inco AI | other | Open weights | | | Inco AI releases DFlash 2 draft model for Qwen3.8-Flash-Next |
| 19 Oct 2026 | Timestral-209M ↗ | Mistral AI | other | Open weights | | 32,768 time steps | Mistral AI releases Timestral forecasting model family, beats Google TimesFM-3 |
| 18 Oct 2026 | DeepSeek-V4.1-Flash-20261017 | DeepSeek | Language | Open weights | MIT | 1M tokens | DeepSeek releases V4.1-Flash-1017 update with lower cost and higher DeepSWE score |
| 18 Oct 2026 | WiFireNet | Swedish Defense Research Agency (FOI) | other | Open weights | | | Swedish FOI unveils WiFireNet AI that detects concealed handguns via Wi‑Fi signals |
| 17 Oct 2026 | REKI ↗ | Kyutai | Multimodal | Open weights | CC BY-NC-SA 4.0 | | Kyutai releases REKI 2.4B vision-language-action model for robot manipulation |
| 17 Oct 2026 | GPT-5.6 Sol | OpenAI | Language | Consumer app | | | OpenAI's GPT-5.6 Sol used unauthorized Sepolia ETH to deploy smart contract |
| 17 Oct 2026 | DiscoveryRSI-27B ↗ | Gen-Verse | Language | Open weights | | | Gen-Verse releases DiscoveryRSI-27B for scientific hypothesis generation |
| 17 Oct 2026 | Qwen3.8-27B-Defiance-Uncensored-MTP-GGUF | HauhauCS | Language | Open weights | Apache-2.0 | 131K tokens | HauhauCS releases uncensored Qwen3.8-27B-Defiance GGUF |
| 17 Oct 2026 | NVIDIA-Nemotron-Labs-Dory1.0-4B-BF16 | NVIDIA | Language | Open weights | OpenMDW-1.1 | 256K tokens | NVIDIA releases Nemotron 3.5 Super VL and Bolt edge models for enterprise agents |
| 16 Oct 2026 | Glimpse | Elorian | Multimodal | API | | | Elorian releases Glimpse, a visual reasoning model with API access |
| 16 Oct 2026 | RTX Spark N1X | Nvidia | other | | | | NVIDIA launches RTX Spark laptops with local AI agents globally |
| 16 Oct 2026 | Beam-1.0-501B-A23B ↗ | Reflection | Language | Open weights | Apache 2.0 | 1M tokens | Reflection releases Beam, a 501B-parameter open-weight model under Apache 2.0 |
| 16 Oct 2026 | Hy-Memory-30B-A3B ↗ | Tencent Hunyuan | Language | Open weights | | | Tencent releases Hy-Memory-30B-A3B for external memory frameworks |
| 16 Oct 2026 | Meta Spark Open | Meta | Language | Open weights | | | Meta launches Meta Spark Open, optimized for Apple Mac Studio |
| 16 Oct 2026 | d1-ultra | Liquid AI | Multimodal | API | | | Liquid AI releases d1-ultra decision model at $0.042 per million tokens |
| 16 Oct 2026 | Nemotron 3.5 Super VL 120B | Nvidia | Multimodal | | | | NVIDIA releases Nemotron 3.5 Super VL and Bolt edge models for enterprise agents |
| 16 Oct 2026 | LitGene ↗ | University of New Mexico | other | Open weights | Creative Commons Attribution 4.0 International | | LitGene transformer integrates text and ontology to improve gene representations |
| 15 Oct 2026 | Muse Spark Open | ServeurpersoCom (via pull request #30616) | Multimodal | Open weights | | 8192 tokens (residual stream) | Meta releases Muse Spark Open, a 816B-parameter multimodal model with 1M-token context |
| 15 Oct 2026 | Ternary Bonsai 2 2B ↗ | PrismML | Language | Open weights | Apache 2.0 | | PrismML releases Ternary Bonsai 2 models for local AI on phones and wearables |
| 15 Oct 2026 | Nemotron 3.5 Bolt 74B-A9B | NVIDIA | Language | Open weights | OpenMDW-1.1 | 256,000 tokens | NVIDIA releases Nemotron 3.5 Super VL and Bolt edge models for enterprise agents |
| 15 Oct 2026 | Nemotron 3.5 Super VL 120B-A12B | Nvidia | Multimodal | Open weights | | | NVIDIA releases Nemotron 3.5 Super VL and Bolt edge models for enterprise agents |
| 15 Oct 2026 | Ling 3.1 flash (quantized GGUF builds) | AtomicChat | Language | Open weights | MIT | 32768 tokens | AtomicChat releases quantized Ling 3.1 flash models with 86–88% token accuracy |
| 15 Oct 2026 | FLUX 3 Image Klein ↗ | Black Forest Labs | Image | Open weights | FLUX Kommunity License | | Black Forest Labs releases FLUX 3 Image Klein, a 5.6B-Parameter Open-Weights model for text-to-image generation |
| 15 Oct 2026 | NVIDIA-Nemotron-3.5-Bolt-74B-A9B-24GB ↗ | NVIDIA | Multimodal | Open weights | OpenMDW License Agreement, version 1.1 | 256K tokens | NVIDIA releases Nemotron 3.5 Super VL and Bolt edge models for enterprise agents |
| 15 Oct 2026 | Apertus 1.5 ↗ | Project Apertus | Multimodal | Open weights | | 262,144 tokens | Apertus 1.5 adds 4T tokens to 8B model, 2T to 70B, with multimodal and long-context upgrades |
| 15 Oct 2026 | Litespark Nemotron 3.5 Super VL | Mindbeam | Language | Open weights | | | Mindbeam announces collaboration with NVIDIA on Litespark Nemotron 3.5 Super VL model and inference engine |
| 15 Oct 2026 | Grok Voice 3.0 | SpaceXAI | Audio | API | | | xAI releases Grok Voice 3.0 for voice agents at $0.08 per minute |
| 15 Oct 2026 | GroundGUI-8B ↗ | PrentisAI | Multimodal | Open weights | | | PrentisAI releases GroundGUI-8B model for GUI agent tasks |
| 15 Oct 2026 | Seal ↗ | Troyanskaya lab / Flatiron Institute | other | Open weights | | | Researchers introduce Seal AI model for gene regulation in developing human brain |
| 14 Oct 2026 | Fish Audio Drama 3 ↗ | Fish Audio | Multimodal | Open weights | Fish Audio Community License v1.0 | | Fish Audio introduces Drama 3 for editing AI voice performances word by word |
| 14 Oct 2026 | Minos | Leash Bio | Language | Open weights | | | Leash Bio releases Minos, an AI playground for lab-in-the-loop simulations |
| 14 Oct 2026 | NVIDIA Nemotron 3.5 Super VL ↗ | NVIDIA | Multimodal | | OpenMDW-1.1 | 1M tokens | NVIDIA releases open-weight Nemotron 3.5 Transcribe model and Granary v2 dataset |
| 14 Oct 2026 | Sarashina ↗ | SB Intuitions | Language | Consumer app | | | SB Intuitions releases Sarashina free beta; translation test shows natural Japanese but meaning shifts |
| 14 Oct 2026 | Grok Voice 3.0 | xAI | Multimodal | API | | | xAI releases Grok Voice 3.0 for voice agents at $0.08 per minute |
| 14 Oct 2026 | Muse Spark Open ↗ | Meta | Multimodal | | Muse Model License | 1M tokens | Meta releases Muse Spark Open, an 816B-parameter open-weight multimodal model |
| 14 Oct 2026 | Neuron AI ↗ | NEURA Robotics | Robotics | | | | NEURA Robotics launches Neuron AI foundation model for Physical AI |
| 14 Oct 2026 | Opus 6 | Anthropic | Multimodal | | | | Anthropic may launch Opus 6 next week to rival OpenAI’s Bell |
| 13 Oct 2026 | Dyna-2 ↗ | Dyna Robotics | Robotics | Consumer app | | | Dyna Robotics deploys Dyna-2.1 robots folding napkins at Din Tai Fung with 99.4% success rate |
| 13 Oct 2026 | Ultra 4K | Meshy | Multimodal | Consumer app | | | Meshy introduces Ultra 4K for 3D models with 80M triangles |
| 13 Oct 2026 | LFM2.5-24B-A2B | Liquid AI | Language | API | | 131,072 tokens | LFM2.5-24B-A2B: Built for fast on-device agents |
| 13 Oct 2026 | DLSS 5 | Nvidia | Image | Consumer app | | | Nvidia speeds DLSS 5 with 2x performance, 20% less VRAM |
| 13 Oct 2026 | Drama 3 | Fish Audio | Multimodal | API | | | Fish Audio introduces Drama 3 for editing AI voice performances word by word |
| 13 Oct 2026 | Wonder | Meta | Language | Consumer app | | | Meta introduces Muse Wonder and Muse Wonder Full frontier models for agents |
| 13 Oct 2026 | Muse Wonder Full ↗ | Meta Superintelligence Labs (MSL) | Language | Consumer app | | | Meta introduces Muse Wonder and Muse Wonder Full frontier models for agents |
| 13 Oct 2026 | GLM-Legal | Z.ai | Language | API | | 1M tokens | Z.ai’s GLM-Legal scores 19.2% on Harvey LAB, outpacing Claude Fable 5 and Kimi K3 |
| 12 Oct 2026 | Muse Wonder (Max) | Meta AI | Language | | | | Meta introduces Muse Wonder and Muse Wonder Full frontier models for agents |
| 12 Oct 2026 | Muse Wonder Full | Meta | Language | Consumer app | | 1M tokens | Meta launches Muse Wonder and Muse Wonder Full agentic AI models |
| 12 Oct 2026 | MiniCPM-V 4.7-1B ↗ | openbmb | Multimodal | Open weights | | | Openbmb releases MiniCPM-V 4.7-35B-A3B, first open-source MoE model for on-device use |
| 12 Oct 2026 | Muse Wonder | Meta’s Superintelligence Lab | Language | API | | | Meta introduces Muse Wonder and Muse Wonder Full frontier models for agents |
| 12 Oct 2026 | MiniCPM-V 4.7-35B-A3B ↗ | openbmb | Multimodal | Open weights | | | Openbmb releases MiniCPM-V 4.7-35B-A3B, first open-source MoE model for on-device use |
| 12 Oct 2026 | Muse Wonder | Meta Superintelligence Lab | Language | API | | | Meta introduces Muse Wonder and Muse Wonder Full frontier models for agents |
| 12 Oct 2026 | Muse Wonder 1.0 ↗ | Meta | Language | API | | 1M tokens | Meta introduces Muse Wonder and Muse Wonder Full frontier models for agents |
| 12 Oct 2026 | NVIDIA-Nemotron-3.5-Super-VL-120B-A12B-BF16 ↗ | NVIDIA | Multimodal | Open weights | OpenMDW License 1.1 | 1M tokens | NVIDIA releases Nemotron 3.5 Super VL 120B with 1M-token context and speculative decoding |
| 11 Oct 2026 | Clef-omni ↗ | Cloudflare | Multimodal | Open weights | Apache-2.0 | 64,000 tokens | Cloudflare releases Clef-omni, a 30B multimodal decision model for text, image, audio and video |
| 11 Oct 2026 | Decision 3.0 | vLLM Semantic Router | Multimodal | Open weights | Apache-2.0 | | vLLM Semantic Router releases Decision 3.0 multimodal models with 5 sizes under Apache-2.0 |
| 11 Oct 2026 | MOSS-Image-SFT ↗ | OpenMOSS | Image | Open weights | Apache License, Version 2.0 | | OpenMOSS releases MOSS-Image-SFT text-to-image model |
| 10 Oct 2026 | Microsoft-Decision-1 | Microsoft | Language | API | | 32,768 tokens | Microsoft releases Decision-1 AI model starting at $0.042 per million input tokens |
| 7 Oct 2026 | Claude Haiku 5.5 | Anthropic | Language | API | | 1M tokens | OpenAI and Anthropic cut AI model prices by up to 80% amid rising competition from open-weight models |
| 7 Oct 2026 | GPT-6 ↗ | OpenAI | Language | Consumer app | | | OpenAI releases 722 math papers and GPT-6 to all users |
| 6 Oct 2026 | EmbeddingGemma 2 | Google DeepMind | Embedding | Open weights | Apache 2.0 | 8K tokens | EmbeddingGemma 2: an open, lightweight multimodal embedding model |
| 6 Oct 2026 | Mistral Large 4 ↗ | Mistral | Multimodal | Open weights | modified MIT | 1M tokens | Mistral releases Mistral Large 4 (Le Chonk) 1-trillion-parameter multimodal model on Hugging Face |
| 6 Oct 2026 | Mistral Large 4 (Le Chonk) | Mistral AI | Multimodal | Open weights | Mistral Ecosystem Open-Weight License (modified MIT) | 1M tokens | Mistral releases Mistral Large 4 (Le Chonk) 1-trillion-parameter multimodal model on Hugging Face |
| 6 Oct 2026 | JEPA-Anything | PhAI Labs | other | Open weights | | | PhAI Labs introduces JEPA-Anything framework for cross-domain world models |
| 5 Oct 2026 | GLM 5.3 | Z.ai | Language | API | | | Z.ai's GLM 5.3 model becomes available on Amazon Bedrock |
| 5 Oct 2026 | Beam 1.0 | Reflection | Language | Open weights | | | Reflection releases Beam, a 501B-parameter open-weight model under Apache 2.0 |
| 5 Oct 2026 | Sona | Yandex | Multimodal | | | | Yandex introduces Sona, a generative recommender that replaces recommendation cascade |
| 5 Oct 2026 | apex-flash-1 | Cantina Security | Code | Open weights | MIT | | Cantina Security releases apex-flash-1, a 321.3B open model for vulnerability research |
| 3 Oct 2026 | Kolibri-1 ↗ | Aleph Alpha | Language | Open weights | Apache 2.0 | 1,048,576 tokens | Aleph Alpha releases Kolibri-1: 78B MoE model with 3.46B active parameters for German and English |
| 3 Oct 2026 | MAI-Transcribe-2-Streaming | Microsoft | Multimodal | API | | | Microsoft launches MAI-Transcribe-2-Streaming real-time transcription AI for $0.54 per hour |
| 3 Oct 2026 | Interfaze 1 Lite ↗ | Interfaze AI | Multimodal | Open weights | | 131k tokens | Interfaze AI releases Interfaze 1 Lite model for developer workloads |
| 2 Oct 2026 | Praxis-1 | Runway AI Inc. | other | Open weights | | | Runway introduces Praxis-1 world action model for robotics |
| 2 Oct 2026 | Clef ↗ | Cloudflare | Multimodal | Open weights | Apache-2.0 | 64,000 tokens | Cloudflare launches Clef decision models to compete with TypeSafe’s Jev |
| 2 Oct 2026 | AstaBrief 8B | Allen Institute for AI (Ai2) | Language | Open weights | Apache 2.0 | | Allen Institute for AI open-sources AstaBrief 8B for faster scientific reports |
| 2 Oct 2026 | Ajax | PewDiePie | Language | | | | PewDiePie releases Ajax, an uncensored Qwen3.5-9B model for home PCs |
| 2 Oct 2026 | Speech | Suno | Audio | Consumer app | | | Suno launches Speech feature for generating voice and music |
| 2 Oct 2026 | Flux 3 Image | Black Forest Labs | Image | API | | | Black Forest Labs releases Flux 3 Image with multi-step editing |
| 2 Oct 2026 | Strands Decider 2B | AWS Strands Labs | other | Open weights | Apache-2.0 | | Amazon releases Strands Decider 2B, a Jev-inspired open-source decision model |
| 2 Oct 2026 | pplx-decider-v1-27b ↗ | Perplexity AI | other | Open weights | | | Perplexity AI releases decision model pplx-decider-v1-27b fine‑tuned from Qwen3.8-27B |
| 1 Oct 2026 | GPT-6 Astra Ultrafast | OpenAI | Language | API | | | NVIDIA Blackwell GPUs power OpenAI's GPT-6 Astra Ultrafast with up to 8x faster token generation |
| 1 Oct 2026 | GPT-6.1 Soul | OpenAI | Language | API | | | OpenAI launches GPT-6.1 Sol at one-fifth of Astra's price |
| 1 Oct 2026 | Strands Decider 2B | Strands Labs | other | Open weights | | | Amazon releases Strands Decider 2B, a Jev-inspired open-source decision model |
| 1 Oct 2026 | Chord | Decagon Labs | Audio | | | | Decagon launches Voice 3 agent with Chord speech model for multilingual calls |
| 1 Oct 2026 | Ideogram 4.5 | Ideogram | Image | API | | | Ideogram introduces Ideogram 4.5 for precise image editing |
| 1 Oct 2026 | Embed 5 Pro | Cohere | Embedding | API | | 128K tokens | Cohere launches Embed 5 Pro and Fast for enterprise search and agents |
| 1 Oct 2026 | Strands Decider 2B | Amazon Web Services (AWS) | Language | Open weights | | | Amazon releases Strands Decider 2B, a Jev-inspired open-source decision model |
| 1 Oct 2026 | Gemini 4 Argon | Alphabet | Language | | | | Google releases Gemini 4 Argon to trusted cyber defenders first |
| 1 Oct 2026 | GPT‑Synopsys | OpenAI & Synopsys | Language | | | | Synopsys unveils Autopilot platform for AI-driven chip design agents |
| 1 Oct 2026 | pplx-embed-v2-context-9b-preview | Perplexity Research | Embedding | Open weights | MIT | | Perplexity and turbopuffer release pplx-embed-v2-context-9b-preview embedding model |
| 30 Sept 2026 | Gemini 4 Argon (High) | Google | Multimodal | API | | 1.0M tokens | Google launches Gemini 4 Argon with 1M token output, claims lead on DeepSWE and cybersecurity benchmarks |
| 30 Sept 2026 | Qwen3.8-27B-pi | Qwen (Tongyi Lab) | Language | Open weights | MIT | | Qwen3.8-27B-pi fixes coding agent effort levels with RL fine-tuning |
| 30 Sept 2026 | Ling-3.1-flash-1019 | inclusionAI | Language | Open weights | | 1M tokens | inclusionAI releases Ling-3.1-flash-1019 with 1M-token window and 560B parameters |
| 30 Sept 2026 | Gemini 4 Argon ↗ | Google | Language | API | | 1 million tokens | Google launches Gemini 4 Argon in Japan with up to 1 million-token context |
| 29 Sept 2026 | d1 ↗ | Liquid AI | Language | API | | | Liquid AI launches d1 decision model for structured tasks with zero output tokens |
| 29 Sept 2026 | GLM-5.3 | Zhipu AI | Language | Open weights | | | Zhipu automates infrastructure with GLM-5.3’s outer RSI loop in under two weeks |
| 29 Sept 2026 | Gemini 3.8 Flash Lite TTS | Google | Audio | Consumer app | | | Google releases Gemini 3.8 Flash TTS for multilingual voice |
| 29 Sept 2026 | Kumo Tabular ↗ | NVIDIA | other | Open weights | OpenMDW-1.1 | | NVIDIA releases Kumo Tabular model for tabular prediction with open weights |
| 29 Sept 2026 | Eleven v4 | ElevenLabs | Multimodal | API | | 10,000 characters | ElevenLabs launches v4 speech model with 10,000-character limit and Turbo variant |
| 29 Sept 2026 | IQuest-Q1 | IQuestLab | Language | API | | 512K tokens | IQuestLab releases IQuest-Q1, a 320B-parameter MoE model for coding agents |
| 29 Sept 2026 | Qwen-Audio-3.1-Realtime | Alibaba Qwen | Multimodal | API | | 262K tokens (245K input, 16K output) | Alibaba releases Qwen-Audio-3.1-Realtime for voice agents at 85% lower prices |
| 29 Sept 2026 | Claude Sonnet 5.5 | Claude | Language | Consumer app | | | Claude Opus 5.5 costs 60% less than Fable 5.1 for writing tasks |
| 29 Sept 2026 | Muse Spark 1.3 | Meta Superintelligence Labs | Multimodal | API | | | Meta's Muse AI agent app reaches #1 on App Store and compiles lists of vulnerable groups on request |
| 29 Sept 2026 | IndustryLLM ↗ | Alibaba Multimodal Industrial AI | Language | Open weights | | | Alibaba releases IndustryLLM for industrial procurement with 35B parameters |
| 28 Sept 2026 | Jeff-Qwen3.5-0.8B ↗ | Independent project (MIT license) | Language | Open weights | Apache 2.0 (weights), MIT (code) | | Open-source decision models Jeff-Qwen3.5-0.8B and Jeff-Gemma4-E2B run in ~30ms on local hardware |
| 28 Sept 2026 | Holo4 ↗ | H company | Language | API | | | H company launches Holo4 agentic models for desktop and API tasks |
| 28 Sept 2026 | Ember-1 | Fireworks AI | Language | API | | | Fireworks AI releases Ember-1, a post-trained Kimi K3 model |
| 28 Sept 2026 | Sonnet 5.5 | Anthropic | Language | API | | | Anthropic cuts AI costs with faster, cheaper Claude Sonnet 5.5 |
| 28 Sept 2026 | Claude Sonnet 5.5 ↗ | Anthropic | Language | API | | 1M tokens | Anthropic cuts costs for Opus 5.5, matching Fable 5.1’s performance |
| 28 Sept 2026 | GPT-6.1 Astra | OpenAI | Language | | | | OpenAI shelves GPT-6.1 Astra after safety concerns about deception and scope |
| 28 Sept 2026 | Apex-2 ↗ | DW-dev-UE | Language | Open weights | Apache 2.0 | 4096 tokens | DW-dev-UE releases Apex-2, a 3.87B MoE LLM trained on 86.5B tokens |
| 27 Sept 2026 | Laya | NandhaKishorM | other | Open weights | Apache-2.0 | | Laya, a 322M-parameter decision engine, gains 26,639 GitHub stars in nine days |
| 27 Sept 2026 | Naive-N0.5-Flash ↗ | NaiveAI | Language | Open weights | MIT | 1M tokens | NaiveAI releases Naive-N0.5-Flash 309B MoE model with 1M context under MIT license |
| 26 Sept 2026 | Opus 5.5 | Anthropic | Language | API | | | Anthropic reports 3.6x jump in token volume after Claude Opus 5.5 launch |
| 26 Sept 2026 | Saaras V4 | Sarvam AI | Multimodal | API | | | Sarvam AI launches Saaras V4 speech-to-text model for 22 Indian languages and global English |
| 26 Sept 2026 | Julia 1 | Supersonic Labs | other | Open weights | Apache 2.0 | 8192 tokens | Supersonic Labs releases Julia 1, a 144.3M-parameter open decision model |
| 26 Sept 2026 | Intern-Decision-4B | InternLM | Multimodal | Open weights | | | Intern-Decision-4B released as multimodal structured decision model |
| 26 Sept 2026 | Pred-MutPRI ↗ | Central China Normal University | other | Open weights | Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International Licen | | Pred-MutPRI predicts mutation-induced binding free energy changes in protein–RNA complexes |
| 25 Sept 2026 | LFM2.5-VL-3B-DSpark | Liquid AI | Multimodal | Open weights | LFM Open License v1.0 | | Liquid AI releases LFM2.5-VL-DSpark for faster vision-language model inference |
| 25 Sept 2026 | Altar-1 ↗ | Aikido Security | Language | Open weights | GLM-5.3 License | | Aikido Security releases Altar-1: a 328 GB security model from GLM-5.3 |
| 25 Sept 2026 | GPT-6 Sol (Max) | OpenAI | Language | API | | | OpenAI announces GPT-6 Astra model |
| 25 Sept 2026 | GLiNER2.5-Decide | Fastino Labs | other | Open weights | Apache 2.0 | | Fastino Releases GLiNER2.5-Decide: A 340M Open-Weight Decision Model That Runs on CPU |
| 24 Sept 2026 | HuatuoGPT-3 ↗ | FreedomIntelligence | Multimodal | Open weights | | | FreedomIntelligence releases HuatuoGPT-3 multimodal model on Hugging Face |
| 24 Sept 2026 | 1-bit Bonsai LLM | PrismML | Language | Consumer app | | | PrismML launches 1-bit Bonsai LLM for Qualcomm smart glasses |
| 24 Sept 2026 | ThinkingCap-Qwen3.8-27B | BottleCap AI | Language | Open weights | PolyForm Small Business 1.0.0 + BottleCap personal-use grant | | BottleCap AI cuts Qwen3.8-27B’s reasoning tokens by 37.2% with ThinkingCap-Qwen3.8-27B |
| 24 Sept 2026 | FLUX 3 Action | Black Forest Labs | Robotics | Open weights | FLUX Kommunity License | | Black Forest Labs releases FLUX 3 Action, an open robotics model |
| 24 Sept 2026 | GPT-6 Cyber | OpenAI | Language | | | | OpenAI may unveil GPT-6 Cyber cybersecurity model in weeks |
| 24 Sept 2026 | LFM2.5-VL-DSpark | Liquid AI | Multimodal | Open weights | | | Liquid AI releases LFM2.5-VL-DSpark for faster vision-language model inference |
| 24 Sept 2026 | Gemini 4 | Google DeepMind | Multimodal | | | | Google DeepMind says Gemini 4 is nearing launch |
| 24 Sept 2026 | CLM-8B | Contrastive-LM | other | | Apache-2.0 | | Contrastive-LM releases CLM-8B, an open model that scores actions up to 9× faster than Jev |
| 24 Sept 2026 | CLM-8B | Stanford University and NVIDIA Research | Language | | | | Contrastive-LM releases CLM-8B, an open model that scores actions up to 9× faster than Jev |
| 23 Sept 2026 | Muse Realtime Avatar | Meta | Multimodal | Consumer app | | | Meta announced the AI device Charm featuring its Muse agent at Connect 2026 |
| 23 Sept 2026 | Gemma 4 26B A4B | Google | Language | Open weights | | | Google Antigravity SDK adds support for local Gemma 4 26B models |
| 23 Sept 2026 | Gemini Omni 1.1 Flash | Google | Video | Consumer app | | | Google expands free access to Gemini Omni 1.1 Flash in Vids with 1080p video generation and duration control |
| 23 Sept 2026 | Nemotron 3 Diarization | Nvidia | Audio | Open weights | OpenMDW License 1.1 | | Nvidia releases free Nemotron 3 Diarization model for real-time speaker identification |
| 23 Sept 2026 | Qwen Image 2.1 | Alibaba Cloud | Image | Open weights | | | Alibaba's Qwen-Image-2.1 claims to beat closed models in image generation |
| 23 Sept 2026 | Qwen-Audio 3.1 ↗ | Alibaba | Audio | API | | | Alibaba cuts Qwen-Audio 3.1 prices by up to 95%, but TTS API is pending |
| 23 Sept 2026 | GPT-6 Sol | Microsoft | Language | API | | | Microsoft adds GPT-6 Sol, Luna and Astra to Foundry model lineup |
| 23 Sept 2026 | GPT-6 Luna ↗ | OpenAI | Language | API | | 1.05 million tokens | OpenAI launches GPT-6.1 Sol at one-fifth of Astra's price |
| 22 Sept 2026 | Jev | Diogo Almeida | other | | | | TypeSafe AI launches Jev, a decision model claimed up to 193.6x faster and 444.6x cheaper |
| 22 Sept 2026 | Gemini 3.8 Flash TTS ↗ | Google | Audio | | | 8K tokens input, 64K tokens output | Google releases Gemini 3.8 Flash TTS for multilingual voice |
| 22 Sept 2026 | GPT-6.1 Sol ↗ | OpenAI | Code | API | | 1,050,000 tokens | OpenAI launches GPT-6.1 Terra for faster, better-performing AI agents at half the price |
| 22 Sept 2026 | Claude Opus 5.5 | Anthropic PBC | Language | API | | | OpenAI announces GPT-6 Astra model |
| 22 Sept 2026 | Apollo | Austrian Academy of Science | Language | Consumer app | | | Austrian Academy of Science to release Apollo, first advanced Greek language model |
| 22 Sept 2026 | Gemini Live | Google | Audio | Consumer app | | | Google launches Gemini 3.8 Live voice model for developers |
| 22 Sept 2026 | Muse ↗ | Meta | other | Consumer app | | | Meta launches Muse AI agent for task automation in Japan and US |
| 22 Sept 2026 | Hy Image 3.5 Preview | Tencent Holdings Ltd. | Image | preview | | | Tencent launches Hy Image 3.5 preview, says it matches ByteDance's Seedream 5.0 Pro |
| 21 Sept 2026 | MiMo-V2.6-Distill-Qwen-9B ↗ | Xiaomi MiMo | Multimodal | Open weights | | | Xiaomi releases MiMo-V2.6-Pro-RL, a 1.02T/42B open omnimodal model trained via scaled RL |
| 21 Sept 2026 | MiMo-V2.6-Flash-RL | Xiaomi | Multimodal | API | | 1M tokens | Xiaomi releases MiMo-V2.6-Pro-RL, a 1.02T/42B open omnimodal model trained via scaled RL |
| 21 Sept 2026 | MiMo-V2.6-Pro | Xiaomi | Language | Open weights | MIT | 1 million tokens | Xiaomi releases MiMo-V2.6-Pro-RL, a 1.02T/42B open omnimodal model trained via scaled RL |
| 21 Sept 2026 | Qwen-Image-2.1 ↗ | Qwen (Alibaba) | Image | Open weights | Qwen Research License Agreement (commercial use requires approval) | | Alibaba releases Qwen-Image-2.1 for open-source image generation and editing |
| 21 Sept 2026 | Claude Opus 5.5 ↗ | Anthropic | Language | API | | 1M tokens | Anthropic files for IPO, reports AI safety incidents and $16.3 billion first-half revenue |
| 21 Sept 2026 | Grok 4.6 ↗ | xAI | Language | API | | 500K token | xAI launches Grok 4.7 for coding and knowledge work at $2 input and $6 output tokens |
| 21 Sept 2026 | Grok 4.7 | xAI | Language | API | | 500K token context window | xAI releases Grok 4.7 on Amazon Bedrock |
| 21 Sept 2026 | Grok 4.7 ↗ | SpaceXAI | Language | API | | 500K token context window | SpaceXAI’s Grok 4.7 now available on Amazon Bedrock |
| 21 Sept 2026 | Grok 4.7 | Grok | Code | API | | | xAI launches Grok 4.7 for coding and knowledge work at $2 input and $6 output tokens |
| 21 Sept 2026 | Step 5 Preview | StepFun | Multimodal | Open weights | | 1M tokens | StepFun releases Step 5 Preview open weights with 600B parameters and 1M context |
| 21 Sept 2026 | Qwen-Image-2.1 | Alibaba Qwen Team | Multimodal | Open weights | | | Alibaba releases Qwen-Image-2.1 for open-source image generation and editing |
| 21 Sept 2026 | TangleDiff | the authors | other | | | | Researchers unveil TangleDiff to design entangled protein hydrogels |
| 20 Sept 2026 | GPT-5.3 Instant | OpenAI | Language | Consumer app | | | Opinion: users complain about preachy ChatGPT, OpenAI tweaks GPT‑5.3 Instant |
| 20 Sept 2026 | Qwen-Image-2.1 | Alibaba | Image | Open weights | Research and evaluation only | | Alibaba releases Qwen-Image-2.1 with transparent background support |
| 20 Sept 2026 | Gander | Tencent | Multimodal | Open weights | | | Tencent unveils Gander model that talks while handling background tasks |
| 20 Sept 2026 | Qwen3.8-LiveTranslate | Alibaba | Multimodal | API | | 53,248 tokens | Alibaba releases Qwen3.8-Omni-Flash, a 1M-token omni-modal model with agentic audio‑video |
| 20 Sept 2026 | Claude Fable 5.1 | Anthropic | Language | API | | | Anthropic cuts cache-reading fees by 75% in Claude Fable 5.1 |
| 20 Sept 2026 | Step-5-Preview-BF16 ↗ | StepFun | Language | Open weights | | | This looks promising: stepfun-ai/Step-5-Preview-BF16 · Hugging Face |
| 19 Sept 2026 | Qwen3.8-Omni-Flash | Qwen | Multimodal | API | | one million tokens | Alibaba releases Qwen3.8-Omni-Flash, a 1M-token omni-modal model with agentic audio‑video |
| 19 Sept 2026 | SPARSEUP | Linkup Research | Embedding | Open weights | Apache 2.0 | | Linkup Research releases sparseup 149M-parameter sparse embedding model |
| 19 Sept 2026 | NASA‑IBM Lunar Foundation Model | NASA and IBM | Multimodal | Open weights | | | NASA and IBM release open-source Lunar Foundation Model for moon science |
| 19 Sept 2026 | Grok Voice Transcribe 2.0 | SpaceXAI | Audio | API | | | xAI releases Grok Voice Transcribe 2.0 speech-to-text model |
| 18 Sept 2026 | Damo Radar | Damo Academy | Multimodal | Open weights | | | Alibaba open-sources Damo Radar model to detect 146 abdominal conditions |
| 18 Sept 2026 | jina-ocr-v1 ↗ | Jina AI | Multimodal | Open weights | CC BY-NC 4.0 | 32,768 tokens | Jina AI releases jina-ocr-v1 3.4B MoE document parser for low‑budget GPUs |
| 18 Sept 2026 | ChatGPT Images 2.5 | OpenAI | Image | API | | | OpenAI releases ChatGPT Images 2.5 with sketch, point, and template tools |
| 18 Sept 2026 | Claude Code 2.1.277 | Anthropic | Code | | | | Anthropic merges Claude chat and Cowork into unified Claude, adds Docs and Slides in beta |
| 18 Sept 2026 | Ternary Bonsai 2 27B | PrismML | Multimodal | Open weights | Apache 2.0 | 262K-token | PrismML hopes its tiny LLM will change how we all use AI |
| 18 Sept 2026 | Grok Voice Transcribe 2.0 | xAI | Audio | API | | | xAI releases Grok Voice Transcribe 2.0 speech-to-text model |
| 18 Sept 2026 | Kimi K3 | Moonshot AI | Multimodal | API | | 1M tokens | Moonshot AI's Kimi K3 opens on Amazon Bedrock with 2.8 trillion parameters |
| 18 Sept 2026 | Jev | OpenAI | Language | | | | OpenAI founder launches Jev model promising lower cost and faster processing |
| 18 Sept 2026 | Qwen3.8-Omni-Flash | Alibaba | Multimodal | API | | 1M tokens | Alibaba releases Qwen3.8-Omni-Flash, a 1M-token omni-modal model with agentic audio‑video |
| 18 Sept 2026 | GPT-6 Astra for Law | OpenAI | Language | API | | | OpenAI launches Astra for Law, a GPT-6 legal AI platform with 230 million‑URL search index |
| 17 Sept 2026 | GPT-6 Astra Law | OpenAI | Language | API | | | OpenAI launches Astra for Law, a GPT-6 legal AI platform with 230 million‑URL search index |
| 17 Sept 2026 | Helix 2.5 | Figure | Robotics | | | | Figure unveils Helix 2.5 humanoid robot, achieving 56% zero-shot success in 30 homes |
| 17 Sept 2026 | HuatuoGPT-3-9B ↗ | FreedomIntelligence | Multimodal | Open weights | | | FreedomIntelligence releases HuatuoGPT-3-9B medical LLM with open usage guides |
| 17 Sept 2026 | Astra for Law ↗ | OpenAI | Language | Consumer app | | | OpenAI releases official ChatGPT add-in for Microsoft Word, available on free plan |
| 16 Sept 2026 | JiT-DDT | Linum | Image | Open weights | Apache 2.0 | | JiT-DDT Architecture Trains Text-to-Image Models 3.6x Faster |
| 16 Sept 2026 | GPT-6 Sol ↗ | OpenAI | Language | API | | 1M tokens | OpenAI launches GPT-6 Sol and Luna with 50% lower API prices |
| 16 Sept 2026 | TabPFN-3.5 | Prior Labs | other | Open weights | | | Prior Labs releases TabPFN-3.5, beating 2015 Kaggle winner with default settings |
| 16 Sept 2026 | LFM2.5-230M-Chess ↗ | mlabonne | Language | Open weights | | | Mlabonne releases LFM2.5-230M-Chess, a 230M-parameter chess engine language model |
| 15 Sept 2026 | Numberwang ↗ | GraafHenk | other | Open weights | MIT | | Zero‑dependency 1.8 MB Numberwang neural net classifies numbers in 11 languages |
| 15 Sept 2026 | Gemini 3.8 Live Extended Thinking | Google DeepMind | Multimodal | Consumer app | | | Google launches Gemini 3.8 Live voice model for developers |
| 15 Sept 2026 | Gemini 3.8 Live ↗ | Google | Multimodal | enterprise | | | Google launches Gemini 3.8 Live with talking AI avatars for enterprise use |
| 15 Sept 2026 | Jev 1.14 | TypeSafe AI | Multimodal | API | | 262k tokens (opt‑in) | TypeSafe AI releases Jev 1.14 with images, box type and 262k token context |
| 15 Sept 2026 | WeatherNext 3 | Google | other | API | | | Google AI Hits 300 Languages, Unveils WeatherNext 3, AlphaGenome Atlas |
| 15 Sept 2026 | Koa | Salesforce | other | Consumer app | | | Salesforce unveils Koa, a reasoning model for its CRM platform |
| 15 Sept 2026 | Claude for Financial Advisors | Anthropic | other | | | | Anthropic releases Claude for Financial Advisors with major wealth tech integrations |
| 13 Sept 2026 | Siri AI | Apple | Multimodal | Consumer app | | | Apple launches iOS 27 with context-aware Siri AI overhaul |
| 12 Sept 2026 | Realtime-Venus-Omni ↗ | inclusionAI | Multimodal | Open weights | Apache License 2.0 | | inclusionAI releases Realtime-Venus 9B audio-visual interaction model |
| 11 Sept 2026 | K2-Horizon-7B-Uno ↗ | Institute of Foundation Models | Language | API | | | IFM releases K2-Horizon-7B-Uno diffusion-augmented LLM, claims 5200 tps speedup |
| 10 Sept 2026 | V4.1 Flash 1017 ↗ | DeepSeek | Multimodal | Open weights | MIT | 1 million tokens | DeepSeek V4.1 Flash 1017 rivals Claude Opus 5 in benchmark tests |
| 8 Sept 2026 | Astra | OpenAI | Language | API | | | OpenAI's GPT-6 Astra: Changing Work and Research |
| 7 Sept 2026 | GPT-6 Astra ↗ | OpenAI | Language | API | | 1.05M tokens | OpenAI adds GPT-6 models to Google Cloud for enterprise buyers |