Benchmark result
GRM-3.2-Sky-ONYX-quality on Intel Graphics — 12.1 tok/s
Measured with llama.cpp manual on October 3, 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
10th fastest of 15 runs of this model on Intel Graphics · median 12.2 tok/s.
sort
- 112.3tok/s80.0ik_llama.cpp:cpu
- 212.3tok/s82.5ik_llama.cpp:cpu
- 312.3tok/s82.7ik_llama.cpp:cpu
- 412.3tok/s78.2ik_llama.cpp:cpu
- 512.2tok/s81.9ik_llama.cpp:cpu
- 612.2tok/s80.2ik_llama.cpp:cpu
- 712.2tok/s82.0ik_llama.cpp:cpu
- 812.2tok/s79.3ik_llama.cpp:cpu
- 912.1tok/s78.6ik_llama.cpp:cpu
- 1012.1tok/s56.4llama.cppthis run
- 1112.0tok/s79.0ik_llama.cpp:cpu
- 128.8tok/s80.7ik_llama.cpp:cpu
- 138.7tok/s79.5ik_llama.cpp:cpu
- 148.7tok/s74.3ik_llama.cpp:cpu
- 158.6tok/s74.8ik_llama.cpp:cpu
Reproduce this run
toolllama.cpp manual
modelGRM-3.2-Sky-ONYX-quality
context262,144
thinkingon
think limitnone
samplingtemperature 0.8500000238418579 · top_p 0.949999988079071 · top_k 40 · min_p 0.05000000074505806 · presence_penalty 0 · repeat_penalty 1.100000023841858
clientv0.4.87+102
Set those in llama.cpp, then:
llm-benchmark benchmark --model "GRM-3.2-Sky-ONYX-quality" --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.