Benchmark result

gemma4:12b-it-qat on NVIDIA GeForce GTX 1080 — 15.1 tok/s

Measured with Ollama 0.33.3 on October 25, 2026.

What the model built

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 GTX 1080 · median 15.2 tok/s.

sort

115.3tok/s77.9Ollama · Q4_0
215.1tok/s69.6Ollama · Q4_0this run

Compare the nearest 5 →

gemma4:12b-it-qat on other hardware

Reproduce this run

toolOllama 0.33.3

modelgemma4:12b-it-qat (Q4_0)

context8,192

thinkingoff

samplingtemperature 1 · top_p 0.95 · top_k 64

clientv0.4.97+102

Set those in Ollama, then:

llm-benchmark benchmark --model "gemma4:12b-it-qat" --tool "Ollama"

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.