Benchmark result
Qwen3.8-MXFP4 on AMD Radeon AI PRO R9700 — 67.6 tok/s
Measured with vLLM manual on October 23, 2026.
What the model built
Open full screen →The coding scenario asks for a playable game in a single HTML file. This is exactly what the model returned, unedited; the frame only adds a line that tells this page how tall it is.
How this run compares
1st fastest of 3 runs of this model on AMD Radeon AI PRO R9700 · median 66.6 tok/s.
sort
- 167.6tok/s84.4vLLMthis run
- 266.6tok/s85.2vLLM
- 364.9tok/s82.1vLLM
Reproduce this run
toolvLLM manual
modelQwen3.8-MXFP4
context8,192
clientv0.4.97+102
Set those in vLLM, then:
llm-benchmark benchmark --model "Qwen3.8-MXFP4" --tool "vLLM"The client prompts for context length and thinking mode, and for the KV cache dtype on Unsloth Studio. Sampling is left at the model default — the values above are what the tool reported using, not overrides the client sent.