@DeRonin_: these guys built an infinite movie generation machine... @fal's post-trained Minimax H3 Max is 50x faster than the orig…
Summary
FastVideo has released an open-source, post-trained version of Minimax H3 that generates video 50x faster, producing 5 seconds of video from 3 seconds of compute for text-to-audio-video generation.
View Cached Full Text
Cached at: 08/29/26, 08:08 PM
these guys built an infinite movie generation machine… @fal’s post-trained Minimax H3 Max is 50x faster than the original: 5 seconds of video out of 3 seconds of compute it generates faster than you can watch it and i found the open-sourced version: https://huggingface.co/FastVideo/FastVideo-FastH3-4-step-Preview-v1-VSA-DataFree…
FastVideo/FastVideo-FastH3-4-step-Preview-v1-VSA-DataFree · Hugging Face
Source: https://huggingface.co/FastVideo/FastVideo-FastH3-4-step-Preview-v1-VSA-DataFree
https://huggingface.co/FastVideo/FastVideo-FastH3-4-step-Preview-v1-VSA-DataFree#fastvideo-fasth3-4-step-preview-v1-vsa-datafreeFastVideo-FastH3-4-step-Preview-v1-VSA-DataFree
The recommended FastH3 Preview v1 checkpoint fromFastVideo. It generates synchronized video and audio from text with four transformer forwards. This step-1300 model was trained with data-free DMD2 and VSA-H3 at 90% sparsity.
Blog·Matching LoRA·FastH3 collection
This checkpoint requires FastVideo’s VSA-H3 attention backend. Use the matching LoRA above if you prefer to download only the distilled adapter.
https://huggingface.co/FastVideo/FastVideo-FastH3-4-step-Preview-v1-VSA-DataFree#run-with-fastvideoRun with FastVideo
Installuv, then use the CUDA 13 / Blackwell path below. It selects FastVideo’s published CUDA kernel wheel instead of compiling the kernel locally. See theinstallation guidefor other platforms.
git clone https://github.com/hao-ai-lab/FastVideo.git
cd FastVideo
uv venv --python 3.12 --seed
source .venv/bin/activate
UV_TORCH_BACKEND=cu130 uv pip install \
--no-sources-package fastvideo-kernel \
-e ".[fasth3]"
python examples/inference/basic/basic_fasth3.py \
--model-path FastVideo/FastVideo-FastH3-4-step-Preview-v1-VSA-DataFree \
--prompt "your prompt" \
--no-warmup \
--repeats 1
The tested defaults use four B200 GPUs and the trained four-forward schedule. On other multi-GPU CUDA systems, follow the installation guide and add\-\-no\-replicated\-dit \-\-vsa\-kernel triton \-\-no\-fa4. The GPU count must divide H3’s 56 attention heads.
https://huggingface.co/FastVideo/FastVideo-FastH3-4-step-Preview-v1-VSA-DataFree#scopeScope
This preview supports text-to-audio-video generation. FL2VA and Ref2VA were not distilled. Difficult motion, fine detail, and some audio may remain below the base MiniMax H3 model. This checkpoint inherits theMiniMax H3 Community License.
https://huggingface.co/FastVideo/FastVideo-FastH3-4-step-Preview-v1-VSA-DataFree#acknowledgementsAcknowledgements
We thankNuva Labfor bringing production grounding to FastH3 through its experience with real-world creative video-agent workloads. Its production-aligned post-training insights help bridge open-source research to practical data-assisted distillation for commercial video workflows, with Omni Ref as the next focus.
We thank theNVIDIA FastGenteam for theDMD2framework and H3 reference experiment that helped us align the score clock, modality shifts, and backward simulation.
We also thankMiniMaxfor releasing H3-Base, and thevLLM project,NVIDIA, andMBZUAIfor their continued sponsorship and support ofFastVideo.
Rehan Sheikh (@rehan_shei): Minimax H3 Max has generates video faster than you can watch it so I hooked it to a twitch livestream! Now you can watch infinite interdimensional cable - link to the stream below
Similar Articles
MiniMax H3 Max (Post-trained by fal on MiniMax H3) sets the new Pareto Frontier for video generation, nearly 50x faster than the base model.
MiniMax H3 Max, post-trained by fal on MiniMax H3, sets a new Pareto Frontier for video generation with generation times nearly 50x faster than the base model, achieving 18x and 24x speed improvements for image-to-video and text-to-video tasks respectively.
MiniMax H3 (10 minute read)
MiniMax launches H3, an open multimodal generation model that handles text, images, video, and audio, generating up to 15 seconds of 2K video with native stereo sound, and plans to open-source the weights.
Introducing H3 Max by fal (5 minute read)
H3 Max is a post-trained version of MiniMax H3 optimized for maximum speed, ranking #1 in human preference evaluations for video quality, prompt understanding, and aesthetics while generating videos up to 35x faster than the official endpoint.
@nazranf_: H3 Max is generating 5 seconds of 768p with native stereo audio in under 3 seconds. That’s faster than real time, and i…
H3 Max is an AI model that generates 5 seconds of 768p video with native stereo audio in under 3 seconds, faster than real-time, and ranks #1 on Design Arena and Artificial Analysis for image-to-video. A demonstration shows its use in continuous video streaming on Twitch.
FastVideo/FastVideo-FastH3-4-step-Preview-v1-VSA-DataFree
FastVideo releases FastH3 Preview v1, an AI model checkpoint that generates synchronized video and audio from text using four transformer forwards, trained with data-free DMD2 and VSA-H3 at 90% sparsity.