FastVideo FastH3 8 Step V2 ComfyUI needs about 15.7 GB of memory at Q4_K_M (a 13.6 GB file) with an 8K context window, so it fits on a 16 GB GPU. This repository has 4 quants, from 10.8 GB (Q3_K_M) to 19.1 GB (Q6_K).
FastVideo FastH3 8 Step V2 – text‑to‑audio‑video generation
FastVideo FastH3 8 Step V2 is a GGUF model that generates synchronized video and audio from text. It uses eight transformer forwards and is trained with data‑free DMD2 and VSA‑H3 at 80% sparsity. The model is a descendant of MiniMax H3 and requires the VSA‑H3 attention backend.
Good at
- Text‑to‑audio‑video generation
- Eight‑step transformer pipeline
- Synchronized audio and video output
Limits
- Requires VSA‑H3 attention backend
- No vision input support
- License is MiniMax H3 Community License
Best for
- Creative video generation
- Audio‑video synchronization tasks
- Rapid prototyping of video content
Who it’s for. Developers and researchers who need a compact text‑to‑audio‑video model for experimentation. The model is suitable for those who can integrate the required VSA‑H3 backend and are working within the MiniMax H3 ecosystem.
Quants & memory
| Quant | File | Est. memory | vs VRAM | Result | |
|---|---|---|---|---|---|
| Q3_K_M | 10.8 GB | 12.6 GB | Too big | Get file → | |
| Q4_K_M | 13.6 GB | 15.7 GB | Too big | Get file → | |
| Q5_K_M | 16.3 GB | 18.7 GB | Too big | Get file → | |
| Q6_K | 19.1 GB | 21.9 GB | Too big | Get file → |
For 8 k context, Q3_K_M needs about 12.6 GB of VRAM, Q4_K_M about 15.7 GB, Q5_K_M about 18.7 GB, and Q6_K about 21.9 GB. Choose Q3_K_M for 8 GB VRAM, Q4_K_M for 12 GB, Q5_K_M for 16 GB, and Q6_K for 20 GB or more.
What GPU do you need for FastVideo FastH3 8 Step V2 ComfyUI?
These figures add the file size, an fp16 KV cache at an 8K context window and about 0.5 GB for the runtime. They are estimates, so leave a little headroom, and expect to use less context or a smaller quant if you are right at the limit.
- 4 GB VRAM: no quant fits at an 8K context.
- 6 GB VRAM: no quant fits at an 8K context.
- 8 GB VRAM: no quant fits at an 8K context.
- 12 GB VRAM: no quant fits at an 8K context.
- 16 GB VRAM: the best quant that fits is Q4_K_M (13.6 GB file, about 15.7 GB in use).
- 24 GB VRAM: the best quant that fits is Q6_K (19.1 GB file, about 21.9 GB in use).
Which FastVideo FastH3 8 Step V2 ComfyUI quant should you choose?
Q4_K_M is the usual starting point for FastVideo FastH3 8 Step V2 ComfyUI: it balances file size and quality well. Here it is a 13.6 GB file.
If memory is tight, the smallest file is Q3_K_M at 10.8 GB, with the quality trade-offs that come with fewer bits. A reasonable choice when a 4-bit file will not fit, mostly on larger models. Quality loss is noticeable but often acceptable.
New to quant names? Read GGUF quantization explained.
Specifications
Architecture from the GGUF header
Source
Frequently asked questions about FastVideo FastH3 8 Step V2 ComfyUI
How much VRAM does FastVideo FastH3 8 Step V2 ComfyUI need?
About 15.7 GB at Q4_K_M with an 8K context window. That is the 13.6 GB file, plus the KV cache and roughly 0.5 GB of runtime overhead. Smaller quants need less and longer context needs more.
Can I run FastVideo FastH3 8 Step V2 ComfyUI on an 8 GB GPU?
Not at an 8K context. The smallest file needs about 12.6 GB, so you would need a larger GPU or CPU offloading.
What is the file size of FastVideo FastH3 8 Step V2 ComfyUI GGUF?
It depends on the quant: Q3_K_M is 10.8 GB, Q4_K_M is 13.6 GB, Q5_K_M is 16.3 GB, Q6_K is 19.1 GB.
Where can I download FastVideo FastH3 8 Step V2 ComfyUI GGUF?
The files are on Hugging Face in the vantagewithai/FastVideo-FastH3-8-Step-V2-ComfyUI-GGUF repository. The download buttons on this page open the file on Hugging Face. ggufmodels.com does not host any model files.
What license does FastVideo FastH3 8 Step V2 ComfyUI use?
Hugging Face lists the license as other. Licenses can differ between the quantized files and the original model, so check the repository and the base model before commercial use.
Can I run FastVideo FastH3 8 Step V2 on 8 GB VRAM?
The Q3_K_M quant requires 12.6 GB VRAM at 8 k context, so 8 GB is insufficient.
Does this model support image or video input?
No, it only supports text‑to‑audio‑video generation.
What license does the model use?
It uses the MiniMax H3 Community License.