Benchmark result
Spark-X2.6-4B-Q8_0 on NVIDIA GeForce RTX 4060 Laptop GPU — 41.3 tok/s
Measured with llama.cpp b11985-6d194531e on October 31, 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
7th fastest of 12 runs of this model on NVIDIA GeForce RTX 4060 Laptop GPU · median 41.7 tok/s.
sort
- 146.2tok/s65.4llama.cpp · Q8_0
- 244.8tok/s54.3llama.cpp · Q8_0
- 344.0tok/s61.2llama.cpp · Q8_0
- 443.3tok/s68.2llama.cpp · Q8_0
- 543.3tok/s68.2llama.cpp · Q8_0
- 642.0tok/s58.2llama.cpp · Q8_0
- 741.3tok/s60.1llama.cpp · Q8_0this run
- 840.8tok/s54.7llama.cpp · Q8_0
- 940.8tok/s66.7llama.cpp · Q8_0
- 1038.0tok/s65.2llama.cpp · Q8_0
- 1137.2tok/s65.1llama.cpp · Q8_0
- 1236.5tok/s71.8llama.cpp · Q8_0
Reproduce this run
toolllama.cpp b11985-6d194531e
modelSpark-X2.6-4B-Q8_0 (Q8_0)
context32,768
thinkingoff
samplingtemperature 0.800000011920929 · top_p 0.949999988079071 · top_k 40 · min_p 0.05000000074505806 · presence_penalty 0 · repeat_penalty 1
clientv0.4.98+102
Set those in llama.cpp, then:
llm-benchmark benchmark --model "Spark-X2.6-4B-Q8_0" --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.