Qwen3.8 Max ships at 2.4T parameters with open weights promised
- Source
- x.com
- Date
Qwen3.8 Max by @Alibaba_Qwen is live on OpenRouter. The new flagship has 2.4T parameters (95B active) and is built for long-horizon coding, research, and multimodal agent work. Open weights are due next week, a first for a Qwen Max-class model. https://t.co/LbRU1pSsW3
Against Qwen3.7 Max, Qwen reports gains across long-horizon and visual agent benchmarks: PaperBench rises from 64.8 to 93.0, OSWorld-Verified from 73.3 to 86.1, and Vision2Web from 42.1 to 69.0.
Context
Hosted access arrived before the promised downloadable release of Alibaba's Qwen3.8 Max. In its August 4 announcement, OpenRouter said developers could already access the model through its service, while the trained model weights were due the following week. That separates two choices for developers: using a remotely hosted model and obtaining the files needed to run it themselves. The post establishes immediate hosted availability and a future commitment on weights; it does not establish that the promised release subsequently occurred.
- multimodal — A model that works with more than text — reading images, audio, or video, and sometimes generating them too.
- AI agent — An AI system that doesn't just answer once but works toward a goal in a loop — taking actions, reading the results, and deciding what to do next.
- open weights — A model whose trained parameters are published for anyone to download and run — unlike API-only models you can access but never possess.
Checking sign-in…
Loading comments…