Benchmark result

qwen3.8-27b-gsq-rco on Apple M5 Max — 29.9 tok/s

Measured with LM Studio on September 11, 2026.

ShareRedditX

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

Reproduce this run

toolLM Studio
modelqwen3.8-27b-gsq-rco
context8,192
clientv0.4.59+97

Set those in LM Studio, then:

llm-benchmark benchmark --model "qwen3.8-27b-gsq-rco" --tool "LM Studio"

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.