Benchmark result

Qwen3.8-27B-ROCMFP2 on AMD Radeon RX 6700/6700 XT/6750 XT / 6800M/6850M XT — 5167.3 tok/s

Measured with vLLM on August 24, 2026.

ShareReddit X

Reproduce this run

toolvLLM

modelQwen3.8-27B-ROCMFP2

context65,536

thinkingon

clientv0.4.40+97

Set those in vLLM, then:

llm-benchmark benchmark --model "Qwen3.8-27B-ROCMFP2" --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.