Benchmark result
gemma-4-26B-A4B-it-qat-UD-Q4_K_XL on Advanced Micro Devices, Inc. [AMD/ATI] HawkPoint1 — 25.0 tok/s
Measured with llama.cpp b12017-5047748f1 on November 15, 2026.
Support llm-bench.io
llm-bench.io is free. If it helped you, I'd appreciate your support to cover the infrastructure costs.
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
6th fastest of 6 runs of this model on Advanced Micro Devices, Inc. [AMD/ATI] HawkPoint1 · median 31.0 tok/s.
- 132.8tok/s70.4llama.cpp · Q4_K
- 232.3tok/s73.2llama.cpp · Q4_K
- 331.7tok/s62.1llama.cpp · Q4_K
- 430.3tok/s68.4llama.cpp · Q4_K
- 530.0tok/s70.5llama.cpp · Q4_K
- 625.0tok/s74.3llama.cpp · Q4_Kthis run
gemma-4-26B-A4B-it-qat-UD-Q4_K_XL on other hardware
Reproduce this run
Set those in llama.cpp, then:
llm-benchmark benchmark --model "gemma-4-26B-A4B-it-qat-UD-Q4_K_XL" --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.