Blueprint by nvidia

Quantize Models to NVFP4 with NVIDIA Model Optimizer

Cut memory ~3.5× vs FP16 while keeping accuracy close to FP8, then validate with an OpenAI-compatible endpoint