RedHatAI/DeepSeek-V4-Flash-NVFP4-FP8

Quantization vs deepseek-ai/DeepSeek-V4-Flash. RedHatAI/DeepSeek-V4-Flash-NVFP4-FP8 is a quantized version of the other (U8/F8_E4M3).

2 other repos hold byte-for-byte identical weight files.

On the Hub · Downloads on Model Pulse