Qwen3.8 27B API: now on FastMetal
Qwen's Qwen3.8 27B (qwen3.8-27b) is available on FastMetal. It is an open-weight, dense vision-language model and the successor to Qwen3.6 27B, which stays in the catalog.
What you get
- Vision and text: on FastMetal it takes image and PDF input alongside text.
- A 262,144-token context window.
- Adjustable thinking depth, from quick replies to deeper reasoning, set per request with
reasoning_effort. - Your data, your choice: the route FastMetal uses for this model does not train on your prompts and runs under zero data retention, so your input is not kept by the host.
- A Japan-hosted version,
japan-qwen3.8-27b, is also in the catalog if you need the model served in Japan (announcement).
Calling it
curl https://api.fastmetal.ai/v1/chat/completions \
-H "Authorization: Bearer $FASTMETAL_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "qwen3.8-27b",
"messages": [{"role": "user", "content": "Hello"}]
}'
Moving from qwen3.6-27b means changing the model value. To try it without code, pick it in chat.
Pricing
Billed per million input and output tokens, with cache reads billed per million tokens too, from a prepaid balance. Current rates are on the model page and pricing. The full list of models is in the catalog.