Benchmark result
gemma4:26b on AMD Radeon RX 7900 XTX — 28.2 tok/s
Measured with Ollama 0.23.1 on May 6, 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
4th fastest of 13 runs of this model on AMD Radeon RX 7900 XTX · median 25.6 tok/s.
sort
- 131.7tok/s64.9Ollama · Q4_K_M
- 229.0tok/s70.7Ollama · Q4_K_M
- 328.6tok/s58.2Ollama · Q4_K_M
- 428.2tok/s63.8Ollama · Q4_K_Mthis run
- 526.5tok/s63.1Ollama · Q4_K_M
- 625.7tok/s68.4Ollama · Q4_K_M
- 725.6tok/s64.6Ollama · Q4_K_M
- 823.7tok/s60.6Ollama · Q4_K_M
- 923.6tok/s63.6Ollama · Q4_K_M
- 1023.5tok/s61.5Ollama · Q4_K_M
- 1122.0tok/s66.9Ollama · Q4_K_M
- 1220.5tok/s44.6Ollama · Q4_K_M
- 1315.9tok/s21.7Ollama · Q4_K_M
Reproduce this run
toolOllama 0.23.1
modelgemma4:26b (Q4_K_M)
context32,768
clientv0.4.20+97
Set those in Ollama, then:
llm-benchmark benchmark --model "gemma4:26b" --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.