FastH3 NVFP4 VAE Decoder

Blackwell NVFP4 video decoders for MiniMax-H3 and FastH3, including FastH3 Omni. They replace the video decoder while retaining the reference encoder, audio VAE and video latent space.

FastVideo · FastH3 Omni decoder setup · FastH3 Omni release collection

Choose a decoder

Tier	Folder	Decoder layers	Resident decoder weights
Quality	`quality-36L/`	36	1.29 GiB
Fast	`fast-8L/`	8	0.31 GiB

Choose the quality tier for closer reconstruction or the fast tier for lower decode cost with a quality tradeoff. Both tiers are lossy.

Use with FastVideo

Download a packed decoder:

hf download FastVideo/FastH3-NVFP4-VAE-Decoder \
  quality-36L/decoder-nvfp4.safetensors --local-dir ./fasth3-vae

Follow the FastVideo FastH3 Omni guide and pass the downloaded file with --decoder to examples/inference/basic/basic_fasth3_omniref_pdd.py. Use fast-8L/decoder-nvfp4.safetensors to select the fast tier instead.

The 36-layer decoder needs a full 36-layer MiniMax-H3 host VAE. The 8-layer tier also works with the light H3 VAE. The FastH3 Omni guide explains model composition and device profiles.

For UniServe deployment of the complete FastH3 Omni model on GB200/B200, RTX PRO 6000 Blackwell or Hopper, follow the UniServe guide with the BF16 model. These files are FastVideo decoder components, not complete diffusion checkpoints.

Hardware

The FP4 decoder kernels require NVIDIA Blackwell: GB200/B200, RTX PRO 6000 Blackwell, RTX 50-series or DGX Spark (GB10). Follow the local FastH3 Omni guide for Spark and RTX 5090 setup. Hopper and older GPUs should use the original H3 video decoder.

Files

Each tier includes:

decoder-nvfp4.safetensors: packed NVFP4 weights for inference.
decoder.safetensors: BF16 master weights.
config.json: decoder architecture and quantization configuration.

Quality and speed

Reconstruction against the original FP32 decoder on the same latents; lower LPIPS and higher PSNR are better:

Evaluation clips	Quality tier LPIPS / PSNR	Fast tier LPIPS / PSNR
FastH3 V2 generations, 9 clips at 480p	0.0104 / 37.9 dB	0.0684 / 28.5 dB
FastH3 Omni generations, 20 clips	0.0039 / 46.0 dB	0.0276 / 36.1 dB
Real videos, VAE-encoded, 6 held-out clips	0.0038 / 45.0 dB	0.0330 / 34.3 dB

Warm video-decode time for one 832×480, 124-frame clip on one GB200, using FastVideo's fused NVFP4 kernels and batching all spatial tiles in one decoder call:

Decoder	Seconds per clip
Original FP16 decoder with `torch.compile`	0.98
Quality, 36 layers	0.53
Fast, 8 layers	0.13

These are decoder-only measurements. Full video/audio generation also includes encoding, denoising and media output.

Local AI resources

Run FastH3 Omni on DGX Spark or RTX 5090 through FastVideo using the FastVideo-format NVFP4 QAD weights. Start with the FastVideo cookbook and FastH3 Omni recipe builder; use the Spark recipe or RTX 5090 recipe and the local inference guide for the AdaLN table, offload and decoder setup.

FP8 weights are a separate FastVideo variant; the FP8 recipe for RTX 4090 illustrates the non-FP4 path with layerwise offload and compatible BF16 encoder components. Use each runtime's matching checkpoint format.

Acknowledgements

FastH3 Omni was built in close collaboration with Nuva Lab, which co-developed the model with FastVideo, defined and curated the omni reference dataset, and brought production-grounded creative video workflows to FastH3. For dedicated production deployment, fine-tuning on your own data and a creative agent on top, see Nuva Lab's FastH3-Omni page.

Thanks also to the NVIDIA FastGen team.