Alibaba released Qwen4-Max on November 16, 2026, which it calls the most capable model in the Qwen family to date. In a press release from Hangzhou, the company said the model has 2.4 trillion total parameters, native vision capabilities and a one-million-token context window. It is a mixture-of-experts design that activates 71 billion parameters per token, which Alibaba says is 25% fewer than Qwen3.8-Max at the same total scale.

Alibaba says Qwen4-Max ranked sixth on Code Arena, a real-world coding leaderboard, and posted strong scores on Agents' Last Exam, Vals Index, JobBench and Chartography. On Work360, a benchmark Alibaba describes as proprietary and covering 721 specialized fields, it says the model outperforms Qwen3.8-Max while using 35% fewer tokens at High reasoning effort and cutting average cost per task by around 60%. The company also described a 48-hour agent run that it says produced an experimentally validated chip-coolant molecule, and a machining showcase it says cut manual programming time from about 475 minutes to 24.

On availability, Alibaba says Qwen4-Max is available now through QwenCloud's Token Plan and API and across Alibaba platforms including Qwen Studio, QwenWork and, according to the coverage, Qoder. The API charges $1.50 per million input tokens and $4.50 per million output tokens in the Singapore region, according to VentureBeat and The Next Web, with cached input at $0.075. Unlike many earlier Qwen releases, Qwen4-Max is proprietary and has no downloadable weights. Alibaba says smaller open-weight Qwen4 models will be released soon but gave no dates or licenses in the material reviewed.

VentureBeat's Carl Franzen, in a November 17 analysis, adds independent context. He reports that Artificial Analysis places Qwen4-Max at 43 on its Intelligence Index, up from 34 for Qwen3.8-Max, but behind Kimi K3.1 Pro at 46, GPT-6.1 Sol and Claude Sonnet 5.5 at 55, Claude Opus 5.5 at 59, Gemini 4 Carbon at 60 and Claude Fable 5.2 at 65. On Artificial Analysis cost per task, he reports $3.27 for Qwen4-Max against $3.68 for Kimi K3.1 Pro, though from a different test suite than Alibaba's own figures. He also notes results that trail rivals, including 48.5 on Terminal-Bench 4.0 against 59.6 for Opus 5.5 and 67.2 for Kimi, and says the model is slow, at roughly 32 minutes per Intelligence Index task in Artificial Analysis testing.

The Next Web reported that a limited preview opened on November 11 and that the Qwen team announced the model on Monday, November 16.

The benchmark charts Alibaba published were not independently rerun, and cost savings from Work360 cannot be assumed to apply to other workloads.

References