Benchmark result
prism-ml/Ternary-Bonsai-2-4B-gguf:PQ2_0 on NVIDIA GeForce RTX 4070 — 214.6 tok/s
Measured with llama.cpp manual on October 16, 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
10th fastest of 13 runs of this model on NVIDIA GeForce RTX 4070 · median 261.1 tok/s.
sort
- 1278.3tok/s54.3llama.cpp · Q2_0
- 2278.0tok/s48.0llama.cpp · Q2_0
- 3276.2tok/s52.0llama.cpp · Q2_0
- 4275.5tok/s48.3llama.cpp · Q2_0
- 5275.0tok/s49.8llama.cpp · Q2_0
- 6272.7tok/s43.4llama.cpp · Q2_0
- 7261.1tok/s56.1llama.cpp · Q2_0
- 8232.2tok/s60.9llama.cpp · Q2_0
- 9228.9tok/s43.2llama.cpp · Q2_0
- 10214.6tok/s49.4llama.cpp · Q2_0this run
- 11212.7tok/s46.9llama.cpp · Q2_0
- 12210.4tok/s54.9llama.cpp · Q2_0
- 13185.0tok/s55.7llama.cpp · Q2_0
Reproduce this run
toolllama.cpp manual
modelprism-ml/Ternary-Bonsai-2-4B-gguf:PQ2_0 (Q2_0)
context262,144
thinkingon
think limitnone
samplingtemperature 1 · top_p 0.949999988079071 · top_k 20 · min_p 0 · presence_penalty 0 · repeat_penalty 1
clientv0.4.96+102
Set those in llama.cpp, then:
llm-benchmark benchmark --model "prism-ml/Ternary-Bonsai-2-4B-gguf:PQ2_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.