Qwen3.8-27B · MXFP4 (FP8 active)
10M output tokens
$9.00
Skip to product information
Qwen3.8-27B · MXFP4 (FP8 active)
$9.00
Fast fully-capable agentic open-weight model allowing for quick interactive development.
- Quality: completes long-horizon agentic coding with or without planning and no looping.
- Context: full 262k supported by the model. Uses prompt caching (no charge for cached input)
- Speed: (Peak) TTFT < 1 sec, 200+ tokens/sec generation; (Typical load) ~80 tokens/sec
- API: OpenAI-compatible
- Early Access Pricing: per 10M output tokens; uncached input counts as 1/10th, cached input is free.
Hugging Face Qwen3.8-27B model card