Benchmark result
prism-ml/Ternary-Bonsai-2-27B-gguf on NVIDIA GeForce RTX 4070 — 45.9 tok/s
Measured with Unsloth Studio on September 27, 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
5th fastest of 9 runs of this model on NVIDIA GeForce RTX 4070 · median 45.9 tok/s.
sort
- 161.7tok/s60.4Unsloth Studio · Q4_K_M · KV q8_0
- 247.9tok/s40.8Unsloth Studio · KV f16
- 347.6tok/s77.9Unsloth Studio · KV f16
- 446.4tok/s53.4Unsloth Studio · TQ1_0 · KV f16
- 545.9tok/s81.7Unsloth Studio · TQ1_0 · KV f16this run
- 645.3tok/s84.5Unsloth Studio · KV f16
- 745.1tok/s72.7Unsloth Studio · TQ1_0 · KV f16
- 843.5tok/s82.8Unsloth Studio · PTQ1_0 · KV q4_0
- 942.9tok/s79.7Unsloth Studio · PQ2_0 · KV q4_0
prism-ml/Ternary-Bonsai-2-27B-gguf on other hardware
Reproduce this run
toolUnsloth Studio
modelprism-ml/Ternary-Bonsai-2-27B-gguf (TQ1_0)
context50,000
kv cachef16
effortxhigh (model default)
think limitnone
samplingtemperature 0.7 · top_p 0.95 · top_k -1 · min_p 0.01 · presence_penalty 0
clientv0.4.80+98
Set those in Unsloth Studio, then:
llm-benchmark benchmark --model "prism-ml/Ternary-Bonsai-2-27B-gguf" --tool "Unsloth Studio"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.