← Back to Blog
New model

Qwen3.8 2.4T-A95B API: the largest Qwen3.8 on FastMetal

FastMetal

Qwen3.8 2.4T-A95B is available on FastMetal as qwen3.8-2.4t-a95b. It joins Qwen3.8 Max and Qwen3.8 27B, which we already serve, and it is the largest model of the three by parameter count.

What you get

  • A 2.4-trillion-parameter mixture-of-experts (MoE) model. Only a subset of experts, about 95B parameters as the name says, runs for each token, so it costs much less to run than its total size suggests.
  • A 1,048,576-token context.
  • A short list of upstream hosts. Some hosts serve this model at lower precision or with a shorter context than the 1M it is built for. We route only to hosts that serve the full context, so you get the same model on every request.

Calling it

Set model to qwen3.8-2.4t-a95b. The base URL and API key are the same as for every other model on FastMetal:

curl https://api.fastmetal.ai/v1/chat/completions \
  -H "Authorization: Bearer $FASTMETAL_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "qwen3.8-2.4t-a95b",
    "messages": [{"role": "user", "content": "List the weak points of this design and propose alternatives."}]
  }'

With the OpenAI SDK, point base_url at https://api.fastmetal.ai/v1 and change model. To try it without code, pick it in chat.

Which Qwen3.8?

The choice comes down to size against cost per token. Qwen3.8 2.4T-A95B vs Qwen3.8 Max shows both models' real answers to the same prompts, and the model catalog lists every Qwen3.8 with its current rates. For how Qwen3.8 Max does against a frontier model, see Qwen3.8 Max vs Claude Opus 5.

Pricing

Billed per million input and output tokens, from a prepaid balance with no monthly fee. Current rates are on pricing and the model page.

Try Qwen3.8 2.4T-A95B right now

Qwen3.8 2.4T-A95B is available on FastMetal through one API key. Start in the browser, or call it from the OpenAI SDK.