Benchmark result
Qwen3.8-27B-oQ4e-fp16-mtp on Apple M5 Max — 30.3 tok/s
Measured with oMLX 0.31.3 on August 20, 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
3rd fastest of 4 runs of this model on Apple M5 Max · median 38.7 tok/s.
sort
- 147.9tok/s85.6oMLX · FP16
- 247.1tok/s90.6oMLX · FP16
- 330.3tok/s78.9oMLX · FP16this run
- 429.1tok/s85.1oMLX · FP16
Reproduce this run
tooloMLX 0.31.3
modelQwen3.8-27B-oQ4e-fp16-mtp (FP16)
context262,144
clientv0.4.39+97
Set those in oMLX, then:
llm-benchmark benchmark --model "Qwen3.8-27B-oQ4e-fp16-mtp" --tool "oMLX"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.