Benchmark result

Qwen3.8-27B-GSQ-RCO-BuffedMod-IQ3_S_XL-mtp-Q6K on 2× NVIDIA GeForce RTX 5060 — 34.6 tok/s

Measured with llama.cpp b11381-836d57176 on October 3, 2026.

What the model built

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

Qwen3.8-27B-GSQ-RCO-BuffedMod-IQ3_S_XL-mtp-Q6K on other hardware

Reproduce this run

toolllama.cpp b11381-836d57176

modelQwen3.8-27B-GSQ-RCO-BuffedMod-IQ3_S_XL-mtp-Q6K

context81,920

thinkingon

think limitnone

samplingtemperature 0.6000000238418579 · top_p 0.949999988079071 · top_k 20 · min_p 0.05000000074505806 · presence_penalty 0 · repeat_penalty 1.0499999523162842

clientv0.4.87+102

Set those in llama.cpp, then:

llm-benchmark benchmark --model "Qwen3.8-27B-GSQ-RCO-BuffedMod-IQ3_S_XL-mtp-Q6K" --tool "llama.cpp"

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.