Benchmark result
qwen3.8-27b-ud-q3_k_xl on NVIDIA GeForce RTX 5080 — 50.8 tok/s
Measured with llama.cpp manual on October 10, 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 5 runs of this model on NVIDIA GeForce RTX 5080 · median 52.8 tok/s.
sort
- 155.1tok/s84.2llama.cpp · Q3_K
- 253.8tok/s83.2llama.cpp · Q3_K
- 352.8tok/s88.5llama.cpp · Q3_K
- 450.8tok/s82.3llama.cpp · Q3_Kthis run
- 549.1tok/s87.4llama.cpp · Q3_K
Reproduce this run
toolllama.cpp manual
modelqwen3.8-27b-ud-q3_k_xl (Q3_K)
context65,536
thinkingon
effortxhigh (model default)
think limit20,000 tokens
samplingtemperature 1 · top_p 0.949999988079071 · top_k 20 · min_p 0.05000000074505806 · presence_penalty 0 · repeat_penalty 1
clientv0.4.89+102
Set those in llama.cpp, then:
llm-benchmark benchmark --model "qwen3.8-27b-ud-q3_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.