Benchmark result
unsloth/Qwen3-30B-A3B-Thinking-2507-GGUF on NVIDIA GeForce RTX 3080 — 26.3 tok/s
Measured with Unsloth Studio on November 7, 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
3rd fastest of 9 runs of this model on NVIDIA GeForce RTX 3080 · median 25.0 tok/s.
sort
- 1199.0tok/s42.6Unsloth Studio · UD-TQ1_0 · KV q4_0
- 233.6tok/s54.1Unsloth Studio · UD-TQ1_0 · KV q8_0
- 326.3tok/s78.0Unsloth Studio · UD-Q5_K_XL · KV q8_0this run
- 426.0tok/s72.4Unsloth Studio · UD-Q5_K_XL · KV q8_0
- 525.0tok/s75.2Unsloth Studio · UD-Q5_K_XL · KV q8_0
- 624.9tok/s72.1Unsloth Studio · UD-Q5_K_XL · KV q8_0
- 723.1tok/s61.0Unsloth Studio · UD-Q5_K_XL · KV q8_0
- 822.6tok/s73.5Unsloth Studio · UD-Q5_K_XL · KV q8_0
- 922.1tok/s71.2Unsloth Studio · UD-Q5_K_XL · KV q8_0
Reproduce this run
toolUnsloth Studio
modelunsloth/Qwen3-30B-A3B-Thinking-2507-GGUF (UD-Q5_K_XL)
context78,000
kv cacheq8_0
think limitnone
samplingtemperature 0.6 · top_p 0.95 · top_k 20 · min_p 0 · presence_penalty 0
clientv0.4.99+102
Set those in Unsloth Studio, then:
llm-benchmark benchmark --model "unsloth/Qwen3-30B-A3B-Thinking-2507-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.