Benchmark result
Swift-Qwen3.8-27b-oQ4e-fp16-mtp on Apple M5 Max — 31.3 tok/s
Measured with oMLX 0.7.0.dev4 on September 23, 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.
How this run compares
3rd fastest of 3 runs of this model on Apple M5 Max · median 44.8 tok/s.
sort
- 148.6tok/s86.7oMLX · FP16 · KV none
- 244.8tok/s85.5oMLX · FP16 · KV none
- 331.3tok/s81.7oMLX · FP16 · KV nonethis run
Reproduce this run
tooloMLX 0.7.0.dev4
modelSwift-Qwen3.8-27b-oQ4e-fp16-mtp (FP16)
context262,144
kv cachenone
effortmedium
samplingtemperature 1 · top_p 0.95 · top_k 0 · repetition_penalty 1
clientv0.4.80+98
Set those in oMLX, then:
llm-benchmark benchmark --model "Swift-Qwen3.8-27b-oQ4e-fp16-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.