Home / AI news / Alibaba releases Qwen4-Max, a 2.4-trillion-parameter model with vision and a 1M-token context
AI news

Alibaba releases Qwen4-Max, a 2.4-trillion-parameter model with vision and a 1M-token context

Alibaba released Qwen4-Max on November 16, 2026, its most capable Qwen model so far: a 2.4-trillion-parameter mixture-of-experts model with native vision and a one-million-token context window, available now through QwenCloud's Token Plan and API, with smaller open-weight Qwen4 models promised soon.

What happened

  • Qwen4-Max has 2.4 trillion total parameters but activates 71 billion per token, which Alibaba says is 25% fewer active parameters than Qwen3.8-Max at the same scale.
  • Alibaba says it ranked 6th on the Code Arena coding leaderboard and, on its own Work360 benchmark, beat Qwen3.8-Max while using 35% fewer tokens at high reasoning effort and cutting average cost per task by around 60%.
  • It is available on QwenCloud through the Token Plan and API and in Alibaba's QwenWork and QwenStudio; Alibaba lists cached input (implicit cache hit) at $0.075 per million tokens.
  • Alibaba says smaller open-weight Qwen4 models will be released to developers worldwide soon, without giving a date.
Why it matters: Qwen4-Max is another frontier-class option for coding and agent work from outside the US labs, and Alibaba is pitching it on lower cost per task. Teams already using Qwen models can test it now; the promised open-weight Qwen4 models matter more for anyone who wants to run models themselves.

Sources

Checked November 16, 2026 against the sources above. Claims about performance or benchmarks are the company's own unless stated otherwise.

More AI news