Model reference · open weights
FastVideo-FastH3-4-step-VSA-Synthetic-Step1900 is an open-weight video model from FastVideo, listed in the AxForge catalogue. AxForge can bring it up on EU-owned hardware for you on request — with the licence handled where one is required.
About
FastVideo-FastH3-4-step-Preview-v1-VSA-Synthetic-Step1900 A FastH3 Preview v1 ablation from FastVideo. It generates synchronized video and audio from text with four transformer forwards. This step-1900 model was trained with DMD2 on synthetic Base H3 videos and uses VSA-H3 at 90% sparsity. Blog · Matching LoRA · Recommended checkpoint · FastH3 collection This checkpoint requires FastVideo's VSA-H3 attention backend. Run with FastVideo Install uv, then use the CUDA 13 / Blackwell path below. It selects FastVideo's published CUDA kernel wheel instead of compiling the kernel locally. See the installation guide for other platforms. 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. Scope 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 the MiniMax H3 Community License. Acknowledgements We thank Nuva Lab for 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 the NVIDIA FastGen team for the DMD2 framework and H3 reference experiment that helped us align the score clock, modality shifts, and backward simulation. We also thank MiniMax for releasing H3-Base, and the vLLM project, NVIDIA, and MBZUAI for their continued sponsorship and support of FastVideo.
Summarised from the published model card. Read the full card on the HuggingFace links below.
Specifications
| Maker | FastVideo |
|---|---|
| Type | Video models |
| Parameters (lead) | 35.0B |
| Variants | 1 |
| Runs with | fastvideo |
| Based on | MiniMaxAI/MiniMax-H3 |
| Released | 2026-08-27 |
| Licence | Commercial licence needed |
How it works
Variants
Open weights ship in several sizes and precisions. One page, all the variants — pick the one that fits your GPU. VRAM figures are estimates from model size.
| Variant | Params | Precision | VRAM | Fits 16 GB | Weights |
|---|---|---|---|---|---|
| FastVideo-FastH3-4-step-Preview-v1-VSA-Synthetic-Step1900 | 35.0B | BF16 | ~80.6 GB | — | Weights ↗ |
Using it via the API
Once AxForge deploys fastvideo-fasth3-4-step-vsa-synthetic-step1900 for you, it answers on the OpenAI-compatible API — the same base URL and keys as every other model. (fastvideo-fasth3-4-step-vsa-synthetic-step1900 below is illustrative; you get the exact model name on deployment.)
$ curl -sS https://api.axforge.ai/v1/videos/generations \
-H "Authorization: Bearer $AXFORGE_API_KEY" \
-H "Content-Type: application/json" \
-d '{"model":"fastvideo-fasth3-4-step-vsa-synthetic-step1900","prompt":"a drone shot over a forest"}'
Licence
The weights are open but its licence needs a commercial agreement for business use. AxForge can arrange that licence and host the model for you — you pay AxForge, we settle with the model’s maker. Ask us for a quote. Read the licence ↗
Sources
Explore