Benchmark result
Qwen3.8-27B-Heretic-Abliterated-Uncensored-GGUF-RVN-Q5_K_M-mtp on NVIDIA RTX A6000 — 29.7 tok/s
Measured with llama.cpp manual on October 26, 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
2nd fastest of 3 runs of this model on NVIDIA RTX A6000 · median 29.7 tok/s.
sort
- 129.7tok/s82.9llama.cpp · Q5_K_M
- 229.7tok/s85.8llama.cpp · Q5_K_Mthis run
- 329.7tok/s84.0llama.cpp · Q5_K_M
Reproduce this run
toolllama.cpp manual
modelQwen3.8-27B-Heretic-Abliterated-Uncensored-GGUF-RVN-Q5_K_M-mtp (Q5_K_M)
context262,144
thinkingon
effortxhigh
think limitnone
samplingtemperature 1 · top_p 0.949999988079071 · top_k 20 · min_p 0 · presence_penalty 0 · repeat_penalty 1
clientv0.4.98+102
Set those in llama.cpp, then:
llm-benchmark benchmark --model "Qwen3.8-27B-Heretic-Abliterated-Uncensored-GGUF-RVN-Q5_K_M-mtp" --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.