Benchmark result
Qwen3.8-Flash-Next-Dynamic-oQ4e-Q4GU-g64-BF16-PLE-MLX-mtp on Apple M4 Max — 49.5 tok/s
Measured with oMLX 0.31.3 on September 13, 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
6th fastest of 14 runs of this model on Apple M4 Max · median 48.8 tok/s.
sort
- 150.8tok/s85.8oMLX · KV none
- 250.1tok/s82.6oMLX · KV none
- 350.0tok/s85.8oMLX · KV none
- 449.9tok/s87.1oMLX · KV none
- 549.7tok/s82.4oMLX · KV none
- 649.5tok/s82.2oMLX · KV nonethis run
- 749.0tok/s75.1oMLX · KV none
- 848.6tok/s87.6oMLX · KV none
- 947.1tok/s83.1oMLX · KV none
- 1046.6tok/s85.7oMLX · KV none
- 1146.0tok/s82.0oMLX · KV none
- 1245.9tok/s78.2oMLX · KV none
- 1345.3tok/s85.0oMLX · KV none
- 1440.0tok/s67.0oMLX · KV none
Reproduce this run
tooloMLX 0.31.3
modelQwen3.8-Flash-Next-Dynamic-oQ4e-Q4GU-g64-BF16-PLE-MLX-mtp
context262,144
kv cachenone
thinkingon
think limitnone
samplingtemperature 1 · 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-Dynamic-oQ4e-Q4GU-g64-BF16-PLE-MLX-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.