Qwen4 Max API: Alibaba's new flagship is live on FastMetal
Qwen4 Max is available on FastMetal as qwen4-max, on the API and in chat. Alibaba released it on November 16, 2026 as the top model of the Qwen4 series and the successor to Qwen3.8 Max.
What's new
- A 2.4 trillion parameter mixture-of-experts model that activates 71 billion parameters per token. Alibaba says that is 25% fewer active parameters than Qwen3.8 Max.
- On Alibaba's own Work360 benchmark, it beats Qwen3.8 Max overall while using 35% fewer tokens at the High reasoning setting. That is Alibaba's figure, not an independent result.
- A 1,000,000-token context window on FastMetal, enough for a large codebase or a full document set in one request.
- It reasons by default, and reasoning tokens count as output tokens. For short questions you can turn reasoning off with the
reasoningparameter. - We confirmed image input, forced tool calls and strict JSON schema output with real calls before launch. PDF attachments are not supported.
Call it
Set model to qwen4-max:
curl https://api.fastmetal.ai/v1/chat/completions \
-H "Authorization: Bearer $FASTMETAL_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "qwen4-max",
"messages": [{"role": "user", "content": "Hello"}]
}'
To try it without code, open chat and pick Qwen4 Max from the model list. You can attach images to a message.
Pricing
Billed per million input and output tokens, with a lower rate for cached input. The current rates are on the model page. FastMetal is prepaid, so you are never charged beyond the balance you loaded.
See real answers first
The model page shows Qwen4 Max's actual answers to FastMetal's shared set of prompts, next to answers from Qwen3.8 Max and other models.
Browse every model in the catalog, or see how billing works on the pricing page.