Benchmark result
Qwen3.8-Flash-Next-Q5GU-Q6Attn-g64-BF16-PLE-MLX-MTP-Q8 on Apple M4 Max — 52.0 tok/s
Measured with oMLX 0.31.3 on September 12, 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
2nd fastest of 3 runs of this model on Apple M4 Max · median 52.0 tok/s.
sort
- 152.3tok/s76.5oMLX · KV none
- 252.0tok/s84.6oMLX · KV nonethis run
- 351.0tok/s87.7oMLX · KV tq8
Reproduce this run
tooloMLX 0.31.3
modelQwen3.8-Flash-Next-Q5GU-Q6Attn-g64-BF16-PLE-MLX-MTP-Q8
context262,144
kv cachenone
effortmedium
think limitnone
samplingtemperature 0.6 · top_p 0.95 · top_k 20 · min_p 0 · repetition_penalty 1
clientv0.4.59+97
Set those in oMLX, then:
llm-benchmark benchmark --model "Qwen3.8-Flash-Next-Q5GU-Q6Attn-g64-BF16-PLE-MLX-MTP-Q8" --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.