Benchmark result
SC117-Swift-IQ3_S on NVIDIA RTX A6000 — 108.6 tok/s
Measured with llama.cpp manual on October 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; the frame only adds a line that tells this page how tall it is.
How this run compares
2nd fastest of 3 runs of this model on NVIDIA RTX A6000 · median 108.6 tok/s.
sort
- 1114.4tok/s85.6llama.cpp
- 2108.6tok/s73.4llama.cppthis run
- 361.8tok/s85.6llama.cpp
Reproduce this run
toolllama.cpp manual
modelSC117-Swift-IQ3_S
context65,536
effortxhigh (model default)
samplingtemperature 1 · top_p 0.95 · top_k 20
clientv0.4.96+102
Set those in llama.cpp, then:
llm-benchmark benchmark --model "SC117-Swift-IQ3_S" --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.