Model Releases
Just dropped 🧑🍳 Fixed version of DeepSeek-V4-Pro-NVFP4 by @NVIDIAAI https://huggingface.co/nvidia/DeepSeek-V4-Pro-NVFP4
NVIDIA has released a corrected version of DeepSeek-V4-Pro-NVFP4, a quantized model variant optimized for NVIDIA hardware using NV-FP4 (NVIDIA's 4-bit floating-point format). The model is available on
NVIDIA has released a corrected version of DeepSeek-V4-Pro-NVFP4, a quantized model variant optimized for NVIDIA hardware using NV-FP4 (NVIDIA's 4-bit floating-point format). The model is available on Hugging Face and represents an improvement over a previous version with bug fixes or optimizations applied. This release likely enables more efficient inference of the DeepSeek V4 model on NVIDIA GPUs while reducing memory requirements through the FP4 quantization format.
Source: Clem Delangue (X) | 2026-05-29