Benchmark result
byteshape/Qwen3.8-27B-GGUF:IQ4_XS-ASCII-Condensed on AMD Radeon RX 7800 XT +1 more — 41.9 tok/s
Measured with llama.cpp b12407-e574a17b0 on November 18, 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
2nd fastest of 3 runs of this model on AMD Radeon RX 7800 XT +1 more · Multi-GPU · median 41.9 tok/s.
sort
- 142.9tok/s87.5llama.cpp
- 241.9tok/s90.4llama.cppthis run
- 341.3tok/s89.0llama.cpp
Reproduce this run
toolllama.cpp b12407-e574a17b0
modelbyteshape/Qwen3.8-27B-GGUF:IQ4_XS-ASCII-Condensed
context65,536
thinkingon
effortxhigh
think limitnone
samplingtemperature 1 · top_p 0.949999988079071 · top_k 20 · min_p 0 · presence_penalty 0 · repeat_penalty 1
clientv0.4.99+102
Set those in llama.cpp, then:
llm-benchmark benchmark --model "byteshape/Qwen3.8-27B-GGUF:IQ4_XS-ASCII-Condensed" --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.