Benchmark result

unsloth/Qwen3.8-27B-GGUF:UD-Q4_K_XL on Advanced Micro Devices, Inc. [AMD/ATI] Device 7551 — 31.2 tok/s

Measured with llama.cpp b11146-7fe450e19 on October 2, 2026.

ShareReddit X

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

3rd fastest of 5 runs of this model on Advanced Micro Devices, Inc. [AMD/ATI] Device 7551 · median 31.2 tok/s.

sort

Compare the nearest 5 →

unsloth/Qwen3.8-27B-GGUF:UD-Q4_K_XL on other hardware

61.5tok/s83.4NVIDIA GeForce RTX 3080 +1 more · Q4_K

Reproduce this run

toolllama.cpp b11146-7fe450e19

modelunsloth/Qwen3.8-27B-GGUF:UD-Q4_K_XL (Q4_K)

context131,072

thinkingon

effortmedium

samplingtemperature 0.699999988079071 · top_p 0.800000011920929 · top_k 20 · min_p 0 · presence_penalty 1.5 · repeat_penalty 1

clientv0.4.85+101

Set those in llama.cpp, then:

llm-benchmark benchmark --model "unsloth/Qwen3.8-27B-GGUF: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.