community pool / model answer

NVIDIA-Nemotron-Nano-12B-v2

chat · nvidia/NVIDIA-Nemotron-Nano-12B-v2

47 tok/s · model median · 18 runs in the model record.

Reference rigQuantizationMedian resultBasisTTFT (ms)Peak VRAMMax contextEngines
RTX 3060 12GB ×24-bit39 tok/smeasured7,7009.9 GB131,072llama.cpp
RTX 3090 24GB4-bit55.8 tok/smeasured3,857- GB131,072llama.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.