VEDA Sparse Attention Models to get x1.7-x2 boost speedcustom node <a target="_blank" rel="ugc" href="https://github.com/veda-sparse/Veda-on-ComfyUI">https://github.com/veda-sparse/Veda-on-ComfyUI</a>. That node also is in Manager- [minimax_h3_t2va_veda_8nfe_600step_preview_fp8.safetensors](<a target="_blank" rel="ugc" href="https://huggingface.co/Veda-Sparse/Minimax-H3-T2VA-Veda-8NFE-600Step-Preview/resolve/main/minimax_h3_t2va_veda_8nfe_600step_preview_fp8.safetensors">https://huggingface.co/Veda-Sparse/Minimax-H3-T2VA-Veda-8NFE-600Step-Preview/resolve/main/minimax_h3_t2va_veda_8nfe_600step_preview_fp8.safetensors</a>)- [minimax_h3_r2va_veda_preview_fp8.safetensors](<a target="_blank" rel="ugc" href="https://huggingface.co/Veda-Sparse/Minimax-H3-R2VA-Veda-Preview/resolve/main/minimax_h3_r2va_veda_preview_fp8.safetensors">https://huggingface.co/Veda-Sparse/Minimax-H3-R2VA-Veda-Preview/resolve/main/minimax_h3_r2va_veda_preview_fp8.safetensors</a>)File save locationComfyUI/│ ├───📂 models/│ │ └─── 📂 veda/│ │ └─── minimax_h3_t2va_veda_8nfe_600step_preview_fp8.safetensors│ │ └─── minimax_h3_r2va_veda_preview_fp8.safetensorsAll models download from <a target="_blank" rel="ugc" href="https://huggingface.co/Comfy-Org/MiniMax-H3/tree/main">https://huggingface.co/Comfy-Org/MiniMax-H3/tree/main</a> Heretic Text Encoder int8- <a target="_blank" rel="ugc" href="https://huggingface.co/ethanfel/Qwen3-VL-32B-Ultra-Heretic-H3-ComfyUI-INT8-ConvRot/tree/main">https://huggingface.co/ethanfel/Qwen3-VL-32B-Ultra-Heretic-H3-ComfyUI-INT8-ConvRot/tree/main</a> Heretic Text Encoder nvfp4 - <a target="_blank" rel="ugc" href="https://huggingface.co/sakamakismile/Qwen3-VL-32B-Heretic-MiniMax-H3-NVFP4/tree/main">https://huggingface.co/sakamakismile/Qwen3-VL-32B-Heretic-MiniMax-H3-NVFP4/tree/main</a> My ComfyUI Build - <a target="_blank" rel="ugc" href="https://huggingface.co/datasets/StefanFalkok/ComfyUI_portable_torch_2.14.0_cu130_cp313_sageattention_triton">https://huggingface.co/datasets/StefanFalkok/ComfyUI_portable_torch_2.14.0_cu130_cp313_sageattention_triton</a>My NSFW TG Channel - <a target="_blank" rel="ugc" href="https://t.me/stefanfalkokmoth">https://t.me/stefanfalkokmoth</a>My TG Channel - <a target="_blank" rel="ugc" href="https://t.me/StefanFalkokAI">https://t.me/StefanFalkokAI</a>My TG Chat - <a target="_blank" rel="ugc" href="https://t.me/+y4R5JybDZcFjMjFi">https://t.me/+y4R5JybDZcFjMjFi</a>

MiniMax H3 Workflows (T2V, I2V, First-Last-Frame, Ref2V) with SageAttention + Turbo Lora + VEDA Attention (x1.7-x2 boost)) + RIFE VFI Interpolation

<ol><li>Added AGSoft Minimax H3 Cache Node to get 20% speed boost without much loss of quality (best cache I've ever used)</li><li>Added VRAM + RAM Cleaning nodes</li></ol>UPD on 08.09.2026<ol><li>Added more references inputs in Ref2video workflow</li><li>Added kitchen attention option</li></ol>UPD on 13.09.2026<ol><li>Added optional workflow for Face Refine</li></ol>

Files with SHA256: 28584d7f8bf80ca1a8e2c7b6b82aafcd3a29d218a74180e0674edfe12db10dc0