Qwen3.8-Flash-Next-FP8

Overview. Qwen3.8-Flash-Next-FP8 is the FP8 release Qwen publishes of Qwen3.8 Flash Next, an open-weight mixture-of-experts model built for fast responses. FP8 is the precision Qwen itself ships for this model, which is why it is listed under that name.

Key strengths. Short time to first token and high output speed, with tool use and structured output. The model supports a context of up to 262,144 tokens; a serving endpoint may set a lower limit.

When to use it. Use it for interactive chat, agents and high-volume tasks where response time matters.

Limitations. No independent benchmark results are in Stav's catalogue yet. The licence is Qwen's own community licence, not Apache 2.0, so check its terms for your use.

Sovereignty note. Qwen is developed by Alibaba (China). The model has open weights and runs on the serving provider's own hardware, so no traffic goes to Alibaba. Where a request is processed depends on the endpoint that serves it.