Understand this story
On November 21, 2026, Qwen4 Max launched on Together AI's Serverless Inference, offering a 1M-token context window, support for text and image inputs, and automatic prompt caching. Together AI's testing showed that the model read all 344 files in the Tornado repository along with a bug report in a single pass, generated a patch 72 seconds later, and passed both a regression test written by the maintainer and the full suite of 1,673 tests; the request contained 652,519 input tokens and cost about 69 cents at the promotional rate. Requests must use streaming, and reasoning is enabled by default, with the associated tokens billed as output tokens. There are currently no other evidence-backed developments.
AI-generated from public reports
Latest development · Report
Together AI Launches Qwen4 Max with Support for a 1M-Token Context WindowDevelopments
1 confirmed developments
- Source:Together AI Blog
Reports and discussion timeline
Browse reports and discussion by time. Their evidence levels remain distinct.
Nov 21
Together AI BlogFirst-partySelected
Together AI Launches Qwen4 Max with Support for a 1M-Token Context Window
Together AI has launched Qwen4 Max on Serverless Inference, supporting a 1M-token context window, text and image inputs, and automatic prompt caching. In testing, the model read all 344 files in the Tornado repository along with the bug report in a single pass, generated a patch after 72 seconds, and passed both a maintainer-written regression test and the full suite of 1,673 tests. The request contained 652,519 input tokens and cost approximately 69 cents at the promotional rate; requests must enable streaming, while thinking is enabled by default and the associated tokens are billed as output tokens.
AI Trends