Benchmark result
cyankiwi/gemma-4-31B-it-AWQ-4bit on NVIDIA GeForce RTX 3080 Ti +1 more — 31.9 tok/s
Measured with sglang manual on October 21, 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
4th fastest of 4 runs of this model on NVIDIA GeForce RTX 3080 Ti +1 more · Multi-GPU · median 32.3 tok/s.
sort
- 132.7tok/s71.5vLLM
- 232.7tok/s78.0vLLM
- 332.0tok/s78.8sglang
- 431.9tok/s64.6sglangthis run
Reproduce this run
toolsglang manual
modelcyankiwi/gemma-4-31B-it-AWQ-4bit
context65,536
thinkingon
clientv0.4.96+102
Set those in sglang, then:
llm-benchmark benchmark --model "cyankiwi/gemma-4-31B-it-AWQ-4bit" --tool "sglang"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.