community pool / model answer

NVIDIA-Nemotron-3-Nano-4B-BF16

chat · nvidia/NVIDIA-Nemotron-3-Nano-4B-BF16

100 tok/s · model median · 64 runs in the model record.

Reference rigQuantizationMedian resultBasisTTFT (ms)Peak VRAMMax contextEngines
RTX 3060 12GB4-bit93.2 tok/smeasured1,080- GB131,072llama.cpp
RTX 3060 Ti 8GB4-bit112.5 tok/smeasured1,8093.9 GB131,072llama.cpp
RTX 3090 24GB4-bit160.3 tok/smeasured1,7090 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.