> ## Documentation Index
> Fetch the complete documentation index at: https://docs.bizyair.vip/llms.txt
> Use this file to discover all available pages before exploring further.

# ComfyUI Environment Updates

> What's new, fixed, and changed in the shared ComfyUI runtime — new capabilities, bug fixes, version bumps, and additions, updates, and removals of custom nodes

**ComfyUI Online Runtime Update: Augest 26,2026**

**Summary:** This update adds 22 custom nodes, updates 2.

**Headline:** MiniMax-H3 video suite (9 nodes) plus Wan/Hunyuan/CogVideoX video tooling, with ComfyUI upgraded to v0.33.0.

## Core Update

* ComfyUI version: `517799b` → `e3a6631` (v0.33.0)
* Effective: immediately (refresh the page)

## New Capabilities

New custom nodes added:

### MiniMax-H3 video suite

* `ComfyUI-MiniMax-H3-Turbo` (commit `55fee86`) — Turbo-accelerated MiniMax-H3 text-to-video.
* `ComfyUI-YCNodes-MiniMax-H3` (commit `1462072`) — YCNodes MiniMax-H3 video generation nodes.
* `ComfyUI_MiniMaxH3_Director` (commit `8418d4f`) — MiniMax-H3 Director-mode video generation.
* `comfyui-minimax-h3-audio-T8` (commit `dc50c2e`) — MiniMax-H3 audio gen + Skin Finish post-processing.
* `comfyui-minimax-h3-blockcache-T8` (commit `28eda98`) — MiniMax-H3 with block cache for faster inference.
* `Comfyui_Minimax_h3_latent_Upscaler` (commit `59f399f`) — Neural latent upscaler for H3, skips VAE decode/encode.
* `ComfyUI-SolAttn_triton` (commit `842c4ea`) — SolAttn triton kernels for faster attention.
* `ComfyUI-MAINodes` (commit `509aefc`) — Contact-Sheet diffusion + Motion Lab (test-time de-roping).
* `ComfyUI-CacheDiT` (commit `1d92bbd`) — Cache-DiT acceleration node.

### Wan / Hunyuan / CogVideoX toolchain

* `ComfyUI-WanVideoStartEndFrames` (commit `340b4f0`) — WanVideo first/last frame control.
* `ComfyUI-wanBlockswap` (commit `5fa2ec0`) — WanVideo blockswap for lower VRAM usage.
* `ComfyUI-WanVideoKsampler` (commit `e92a703`) — WanVideoKsampler node (companion to wanBlockswap).
* `ComfyUI-HunyuanVideoMultiLora` (commit `9e18b97`) — Multi-LoRA loading for HunyuanVideo.
* `ComfyUI-CogVideoXWrapper` (commit `fdb8abd`) — CogVideoX video generation wrapper.
* `ComfyUI-CogVideoX-MZ` (commit `ecdaefe`) — CogVideoX-MZ video generation.
* `ComfyUI-PainterI2VforKJ` (commit `56f6e8a`) — PainterI2V image-to-video.
* `ComfyUI-MochiEdit` (commit `5929b50`) — Edit videos with the Mochi model.

### Motion / portrait / prompt toolchain

* `comfyui-liveportrait` (commit `ffde131`) — LivePortrait portrait animation (KwaiVGI).
* `ComfyUI-HY-Motion1` (commit `1342cd8`) — HY-Motion1 video motion control.
* `ComfyUI-mesh2motion` (commit `11fe6b7`) — mesh2motion 3D motion transfer.
* `ComfyUI-SCAIL2-Easy` (commit `f4ab426`) — SCAIL-2 simplified nodes.
* `comfyui-dynamicprompts` (commit `3f2fff3`) — Dynamic prompt generation (random/combinatorial/Jinja).

## Bug Fixes

* Fixed onnxruntime
* Fixed ComfyUI-LTXVideo runtime issue (commit `531512f` → `15d09ab`)
* Fixed comfyui-minimax-h3-blockcache-T8 runtime issue
* Fixed LoraLoaderBypassModelOnly configuration
* Repaired comfy-kitchen (`0.2.26` → `0.2.31`)
* Upgraded SageAttention to `2.2.0` for compatibility

## Breaking Changes

Updated custom nodes:

* `ComfyUI-KJNodes` `a8fd39c` → `3f20054`
* `ComfyUI-LTXVideo` `531512f` → `15d09ab`