Model launches gather pace across a widening field of labs
AI model release activity expanded sharply across the months tracked by LLM Releases. September 2026 recorded 61 events across 39 labs, including 51 releases. October 2026 followed with 35 events, 22 labs and 27 releases, while November 2026 lists 50 events across 34 labs, including 41 releases.
Recent launches include Alibaba’s Qwen4-Max, Tencent’s Hy4 flash, Google’s Gemini 4 Hydrogen and Anthropic’s Claude Fable 5.2. Open-weight availability also features prominently, with Thinking Machines opening Inkling 2 Small weights under Apache 2.0 and Meituan open-sourcing LongCat-2.5 under MIT. Specialized systems cover translation, coding, legal work, life-sciences research and on-device deployment.
The lifecycle record extends beyond launches to previews, updates, license changes, deprecations, retirements and withdrawals. It includes U.S. government restrictions on Anthropic models and subsequent access restorations, alongside OpenAI previews conducted under government review. Dates follow official announcements or model cards rather than the time entries were added, with each event linked to a primary source.