Benchmark result
maritaca-ai_S5-nano-Q4_K_M on NVIDIA GeForce RTX 3060 — 72.3 tok/s
Measured with llama.cpp b12015-a34de8618 on November 8, 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 2 runs of this model on NVIDIA GeForce RTX 3060 · median 85.2 tok/s.
sort
- 198.2tok/s56.4llama.cpp · Q4_K_M
- 272.3tok/s62.8llama.cpp · Q4_K_Mthis run
Reproduce this run
toolllama.cpp b12015-a34de8618
modelmaritaca-ai_S5-nano-Q4_K_M (Q4_K_M)
context32,768
thinkingon
think limit21,845 tokens
samplingtemperature 1 · top_p 0.949999988079071 · top_k 20 · min_p 0 · presence_penalty 1.5 · repeat_penalty 1
clientv0.4.99+102
Set those in llama.cpp, then:
llm-benchmark benchmark --model "maritaca-ai_S5-nano-Q4_K_M" --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.