Benchmark result
unsloth/Qwen3.8-27B-GGUF:UD-Q4_K_M on Apple M5 Max — 65.6 tok/s
Measured with vLLM on September 27, 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
3rd fastest of 3 runs of this model on Apple M5 Max · median 79.8 tok/s.
sort
- 186.3tok/s76.5splash · Q4_K_M
- 279.8tok/s89.7splash · Q4_K_M
- 365.6tok/s46.6vLLM · Q4_K_Mthis run
Reproduce this run
toolvLLM
modelunsloth/Qwen3.8-27B-GGUF:UD-Q4_K_M (Q4_K_M)
context8,192
clientv0.4.80+98
Set those in vLLM, then:
llm-benchmark benchmark --model "unsloth/Qwen3.8-27B-GGUF:UD-Q4_K_M" --tool "vLLM"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.