🎞️ this is a place for Minimax H3 Video workflows that are working in low-VRAM GPUs
Turbo LORA / Nodes from: https://huggingface.co/larryvrh/MiniMax-H3-Turbo-Lora
or Lightx2v LORA: https://huggingface.co/lightx2v/Minimax-h3-Turbo/tree/main
Audio VAE from: https://huggingface.co/dummy9996/minimax_h3_vae_fp8/tree/main
Video VAE from: https://huggingface.co/Kijai/MiniMax-H3-experimental/tree/main
Uncensored TE from: https://huggingface.co/ethanfel/Qwen3-VL-32B-Ultra-Heretic-H3-ComfyUI-INT8-ConvRot/tree/main
Also you can Download Resources from any of these repos:
https://huggingface.co/tsolful/Minimax_H3_INT4MixedConvRot/tree/main
https://huggingface.co/Abiray/Minimax-H3-nvfp4-INT4-INT8-Convrot/tree/main
https://huggingface.co/Merserk/MiniMax-H3-INT4-ConvRot/tree/main
and ...
Official ComfyUI Minimax H3: https://huggingface.co/Comfy-Org/MiniMax-H3/tree/main
For live previews node get: https://huggingface.co/Kijai/MiniMax-H3-TAE
P.S.: you should learn how to use this model framework - it is really worth your tears and wasted hours of trial/errors ;)
Description
8 Steps + Comfy Attention = Pure Coolness <3
FAQ
Comments (11)
hi. you wrote workflow works on low vram. Uncensored TE that you used in workflow is 25 gb . does it work on 16vram +32 ram?. thank you
Comfyui is very good at managing those numbers on low vram - do not be afraid!!!
8-12 Gb are ok to use this WF!!!
@0l1v1aR0551 tank you for answering
@faryarjy100 I'm always (almost) answering!
Two Three questions:
1) do we use the heretic TE "normally"? How can the bf16 version be so small?
2) What's the difference between the workflow? I assume we should use the most recent one.
1) grab any TE from links provided in WF card;
1.5) test them vigorously for your purposes!
2) workflows are gradually perfected from older to newer one;
2.5) the older one also works - it is provided for easier understanding for newbies ...
@0l1v1aR0551 Thanks! To be honest, I'm running this on a Ryzen AI 370 and the only TE I was able to get to run with H3 are the gguf ones (after adding the matching mmproj file 😅)
The safetensors versions are practically all FP4 and they give me errors (even if according to online posts they should work on AMD, so I'm a bit perplexed... I'll investigate more 😂 )
@TooLittleVRAM yes, INT8 are only for NVIDIA - so you have to use loaders that are best for your system
Thank you so much for your creativity—the workflow is going very smoothly for me.
great to hear! I will be updating it soon!
I've never tried a "Quick Export" template before, so thank you from the bottom of my heart for your ongoing creativity.