Benchmark result
bottlecapai/ThinkingCap-Qwen3.8-27B-GGUF on AMD Radeon RX 7900 XTX — 44.1 tok/s
Measured with Unsloth Studio on September 24, 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.
How this run compares
1st fastest of 5 runs of this model on AMD Radeon RX 7900 XTX · median 36.1 tok/s.
sort
- 144.1tok/s84.1Unsloth Studio · Q4_K_M · KV f16this run
- 242.6tok/s86.4Unsloth Studio · Q4_K_M · KV f16
- 336.1tok/s83.6Unsloth Studio · Q4_K_M · KV f16
- 434.0tok/s84.7Unsloth Studio · Q4_K_M · KV f16
- 532.4tok/s86.2Unsloth Studio · Q4_K_M · KV f16
Reproduce this run
toolUnsloth Studio
modelbottlecapai/ThinkingCap-Qwen3.8-27B-GGUF (Q4_K_M)
context65,536
kv cachef16 (assumed default)
thinkingon
effortxhigh (model default)
samplingtemperature 0.7 · top_p 0.8 · top_k 20 · min_p 0 · presence_penalty 1.5
clientv0.4.80+98
Set those in Unsloth Studio, then:
llm-benchmark benchmark --model "bottlecapai/ThinkingCap-Qwen3.8-27B-GGUF" --tool "Unsloth Studio"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.