[!IMPORTANT] This quantization was uploaded on 2026-05-05 and replaces a previous version. If you downloaded this model before this date, please re-download for the updated weights.

NVIDIA-Nemotron-3-Nano-4B-oQ8-fp16

Quantization details

• Model type: nemotron_h
• Bits: 8
• Group size: 64
• Format: MLX safetensors