Benchmark result
mtplx-qwen38-27b-optimized-quality on Apple M5 Max — 34.5 tok/s
Measured with vLLM on September 1, 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.
How this run compares
2nd fastest of 3 runs of this model on Apple M5 Max · median 34.5 tok/s.
sort
- 137.8tok/s86.8vLLM
- 234.5tok/s86.0vLLMthis run
- 332.8tok/s88.9vLLM
mtplx-qwen38-27b-optimized-quality on other hardware
Reproduce this run
toolvLLM
modelmtplx-qwen38-27b-optimized-quality
context128,000
thinkingon
clientv0.4.46+97
Set those in vLLM, then:
llm-benchmark benchmark --model "mtplx-qwen38-27b-optimized-quality" --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.