Benchmark result
Qwen3.8-27B-8bit on Apple M5 Ultra — 88.4 tok/s
Measured with oMLX 0.7.3 on November 2, 2026.
Support llm-bench.io
llm-bench.io is free. If it helped you, I'd appreciate your support to cover the infrastructure costs.
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 2 runs of this model on Apple M5 Ultra · median 82.9 tok/s.
sort
- 188.4tok/s84.2oMLX · 8bit · KV nonethis run
- 277.4tok/s88.6oMLX · 8bit · KV none
Qwen3.8-27B-8bit on other hardware
Reproduce this run
tooloMLX 0.7.3
modelQwen3.8-27B-8bit (8bit)
context262,144
kv cachenone
thinkingon
effortmedium
think limitnone
samplingtemperature 1 · top_p 0.95 · top_k 20 · repetition_penalty 1
clientv0.4.98+102
Set those in oMLX, then:
llm-benchmark benchmark --model "Qwen3.8-27B-8bit" --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.