Alibaba Cloud releases Qwen4-Max
Alibaba Cloud lists Qwen4-Max as released on November 16, 2026. The flagship multimodal model has a one-million-token context and is available through pay-as-you-go APIs and Token Plan.
What happened
Event details
The official release log lists qwen4-max on November 16, 2026 (Beijing date). Qwen positions it as its next-generation flagship for reasoning, coding, office work, frontend design, and agentic tasks, with text, image, and video input. The context window is 1,000,000 tokens. Maximum output is 131,072 tokens without thinking, or 262,144 tokens for thinking and the final answer combined. Features include streaming, tool calling, structured output, explicit caching, and reasoning summaries. The released model supports pay-as-you-go APIs and Token Plan Personal and Team. It is distinct from qwen4-max-preview, listed on November 11 with Token Plan access only. Preview access restrictions and output limits should not be applied to the released model. The event uses the official calendar date without inferring an exact launch time.
Assessment
Why it matters
A new Qwen flagship release for multimodal reasoning, coding, and agents, with pay-as-you-go API access beyond the earlier Token Plan preview.
Availability
Access notes
qwen4-max is available through pay-as-you-go APIs and Token Plan; qwen4-max-preview remains a separate model ID.