Benchmark result
gemma4:26b-a4b-it-qat on Apple M4 Pro — 45.3 tok/s
Measured with Ollama 0.32.5 on July 29, 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 5 runs of this model on Apple M4 Pro · median 53.4 tok/s.
sort
- 162.3tok/s68.8Ollama · Q4_0
- 262.2tok/s71.3Ollama · Q4_0
- 353.4tok/s72.6Ollama · Q4_0
- 446.0tok/s70.7Ollama · Q4_0
- 545.3tok/s74.0Ollama · Q4_0this run
Reproduce this run
toolOllama 0.32.5
modelgemma4:26b-a4b-it-qat (Q4_0)
context524,288
clientv0.4.30+97
Set those in Ollama, then:
llm-benchmark benchmark --model "gemma4:26b-a4b-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.