Together AI launches Qwen4 Max, demonstrates one-shot fix for Tornado bug
Understand this story
Together AI has launched Alibaba's flagship model Qwen4 Max and demonstrated using it to fix a bug in Tornado: by inputting all 344 files and 652,519 tokens at once, the model produced a patch within 72 seconds, and the patch passed the regression test written by the maintainer for that bug, at a cost of about 69 cents per request. On Together AI, Qwen4 Max currently supports only streaming requests and is offered only in serverless mode, and cannot be used for fine-tuning; thinking mode is enabled by default, with results returned in the reasoning field before the answer. The model enjoys a 30% discount until November 22, 2026.
AI overview based on source reports · Updated 3 hr ago
Coverage timeline
Follow the coverage to explore different perspectives.
- Together AI BlogFeaturedQwen4 Max is live on Together AI
Together AI describes a test in which Qwen4 Max received all 344 files of the Tornado repository (652,519 tokens) plus a bug report in a single tool-free request, and returned a patch in 72 seconds that passes the maintainers' regression tests and Tornado's full 1,673-test suite. The model also cited a precedent in tornado/iostream.py that the bug report never mentioned and that Tornado's maintainers referenced in their own fix. Qwen4 Max is available serverless as Qwen/Qwen4-Max with a 1M-token context window, streaming-only requests, and 30% off through November 22, 2026.
Activity for this story
Not enough continuous observations to show a trend yet.