community pool / model answer

Ministral-3-3B-Instruct-2512

chat · mistralai/Ministral-3-3B-Instruct-2512

73.2 tok/s · model median · 29 runs in the model record.

Reference rigQuantizationMedian resultBasisTTFT (ms)Peak VRAMMax contextEngines
GTX 1080 Ti 11GB4-bit94.9 tok/sreported-5.3 GB8,192llama.cpp
RTX 3060 12GB4-bit73.2 tok/smeasured721- GB131,072llama.cpp
RTX 3060 12GB8-bit78 tok/sreported9905.7 GB4,608llama.cpp

Context tested

The table above is the tested context for each exact rig and quantization cell. Empty fields remain empty rather than being filled with an estimate.