Benchmark result
Qwen3.8-Flash-Next-oQ4e-mtp on Apple M5 Max — 27.9 tok/s
Measured with oMLX 0.31.4.dev132+g94cdcae13 on October 6, 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
1st fastest of 4 runs of this model on Apple M5 Max · median 23.7 tok/s.
sort
- 127.9tok/s81.1oMLX · oQ4e · KV none · 65% of experts in memorythis run
- 223.7tok/s84.8oMLX · oQ4e · KV none · 65% of experts in memory
- 323.6tok/s77.0oMLX · oQ4e · KV none · 65% of experts in memory
- 423.3tok/s83.0oMLX · oQ4e · KV none · 65% of experts in memory
Reproduce this run
tooloMLX 0.31.4.dev132+g94cdcae13
modelQwen3.8-Flash-Next-oQ4e-mtp (oQ4e)
context100,000
kv cachenone
thinkingon
effortxhigh
think limitnone
samplingtemperature 1 · top_p 0.95 · top_k 20 · repetition_penalty 1
clientv0.4.87+102
Set those in oMLX, then:
llm-benchmark benchmark --model "Qwen3.8-Flash-Next-oQ4e-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.