Benchmark result
unsloth--gemma-4-31b-it-UD-MLX-4bit on Apple M5 Max — 16.2 tok/s
Measured with oMLX 0.32.3.dev8+g0d4d3d04f on November 1, 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
2nd fastest of 2 runs of this model on Apple M5 Max · median 18.6 tok/s.
sort
- 121.0tok/s75.7oMLX · 4bit · KV none
- 216.2tok/s73.4oMLX · 4bit · KV nonethis run
Reproduce this run
tooloMLX 0.32.3.dev8+g0d4d3d04f
modelunsloth--gemma-4-31b-it-UD-MLX-4bit (4bit)
context65,536
kv cachenone
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 "unsloth--gemma-4-31b-it-UD-MLX-4bit" --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.