Benchmark result
Ministral-3-3B-Instruct-2512-GGUF-Q5_K_M on NVIDIA GeForce RTX 2050 — 17.8 tok/s
Measured with Lemonade manual on November 20, 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
3rd fastest of 3 runs of this model on NVIDIA GeForce RTX 2050 · median 18.0 tok/s.
sort
- 118.2tok/s51.1Lemonade · Q5_K_M
- 218.0tok/s54.1Lemonade · Q5_K_M
- 317.8tok/s53.6Lemonade · Q5_K_Mthis run
Reproduce this run
toolLemonade manual
modelMinistral-3-3B-Instruct-2512-GGUF-Q5_K_M (Q5_K_M)
context32,768
clientv0.4.99+102
Set those in Lemonade, then:
llm-benchmark benchmark --model "Ministral-3-3B-Instruct-2512-GGUF-Q5_K_M" --tool "Lemonade"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.