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

# What's new in ComfyUI: recent changes

> What changed recently in ComfyUI: ComfyUI releases, newest first, from the ComfyUI changelog.

Product: local
Updated: 2026-11-21

What's new in ComfyUI: ComfyUI releases, newest first, from the [ComfyUI changelog](https://docs.comfy.org/changelog.md).

### v0.46.0 (November 18, 2026)

**Partner Node Updates**

* [**Kling 4.0 aspect ratios**](https://github.com/Comfy-Org/ComfyUI/pull/17815): 4:3 and 3:4 added to the Kling 4.0 nodes
* [**Beeble and Higgsfield on Comfy Router**](https://github.com/Comfy-Org/ComfyUI/pull/17794): SwitchX and Genjutsu motion transfer moved to the Router backend
* [**Claude Sonnet 4.5 removed**](https://github.com/Comfy-Org/ComfyUI/pull/17819): Retired from the Anthropic Claude node
* [**Fish Audio speech-to-text**](https://github.com/Comfy-Org/ComfyUI/pull/17828): Now runs transcribe-1-pro

**New Open-Source Model Support**

* [**PAN 2**](https://github.com/Comfy-Org/ComfyUI/pull/17755): Video generation model support in core

**New Nodes**

* [**Video Deflicker**](https://github.com/Comfy-Org/ComfyUI/pull/17754): Removes flicker from video, with a faster RAFT optical flow
* [**Split Image Channels**](https://github.com/Comfy-Org/ComfyUI/pull/13548): Red, green, blue and alpha outputs from one image

**Improvements**

* [**SenseNova**](https://github.com/Comfy-Org/ComfyUI/pull/16032): Thinking, interleaved generation and self-correction, plus the new ModelSamplingSenseNova node
* [**Qwen3-1.7B**](https://github.com/Comfy-Org/ComfyUI/pull/17787): Supported in Generate Text
* [**Save Image Advanced**](https://github.com/Comfy-Org/ComfyUI/pull/17727): Saves single channel PNGs
* [**Create Camera Info**](https://github.com/Comfy-Org/ComfyUI/pull/17730): Mesh input to frame any mesh
* [**Concatenate Video**](https://github.com/Comfy-Org/ComfyUI/pull/17621): Clips conform to the first clip's frame rate and audio
* [**--force-output-rerun**](https://github.com/Comfy-Org/ComfyUI/pull/17809): Output nodes write their files again on a cache hit
* [**Vocabulary-pruned Qwen checkpoints**](https://github.com/Comfy-Org/ComfyUI/pull/17633): Embedded tokenizers supported

**Bug Fixes**

* [**Dynamic VRAM**](https://github.com/Comfy-Org/ComfyUI/pull/16491): Fixed a sampling schedule leaking between models
* [**APG**](https://github.com/Comfy-Org/ComfyUI/pull/17741): Fixed output changes at cfg 1.0
* [**Google Omni**](https://github.com/Comfy-Org/ComfyUI/pull/17785): Video is downloaded when Google marks its file as failed

**Removals**

* [**DirectML support removed**](https://github.com/Comfy-Org/ComfyUI/pull/17711): The --directml flag is gone, use CUDA or CPU

### v0.45.6 (November 17, 2026)

**Partner Node Updates**

* [**Kling 4.0 Image and Long Image**](https://github.com/Comfy-Org/ComfyUI/pull/17797): Image generation and editing with up to 10 reference images, new long webpage and e-commerce layouts
* [**HeyGen Avatar 6**](https://github.com/Comfy-Org/ComfyUI/pull/17793): Reference to Video and Identity to Video at 1080p and 2K

### v0.45.5 (November 17, 2026)

**Partner Node Updates**

* [**Bria RMBG 3.0**](https://github.com/Comfy-Org/ComfyUI/pull/17780): RMBG 3.0 model option on the Bria Remove Image Background node

### v0.45.4 (November 15, 2026)

**Partner Node Updates**

* [**Seedance 2.5 Flash**](https://github.com/Comfy-Org/ComfyUI/pull/17764): New Seedance 2.5 model at 480p and 720p
* [**Meta Muse Wonder, Spark 1.3 and Ember**](https://github.com/Comfy-Org/ComfyUI/pull/17744): Three Meta models added to OpenRouter LLM
* [**Claude Fable 5.2**](https://github.com/Comfy-Org/ComfyUI/pull/17738): New model on the Anthropic Claude node
* [**HeyGen Video 1.0**](https://github.com/Comfy-Org/ComfyUI/pull/17721): 1080p output, 30-second clips and auto duration
* [**Krea on Comfy Router**](https://github.com/Comfy-Org/ComfyUI/pull/17691): Krea nodes moved to the Router backend
* [**Wan on Comfy Router**](https://github.com/Comfy-Org/ComfyUI/pull/17639): Wan nodes moved to the Router backend
* [**Luma legacy nodes removed**](https://github.com/Comfy-Org/ComfyUI/pull/17714): Legacy nodes retired, Ray 3.2 and UNI-1 remain
* [**Qwen3.6 Max Preview removed**](https://github.com/Comfy-Org/ComfyUI/pull/17737): Retired from the OpenRouter LLM node

**Bug Fixes**

* [**Fish Audio Drama 3**](https://github.com/Comfy-Org/ComfyUI/pull/17718): Full price shown in the badge
* [**Topaz video nodes**](https://github.com/Comfy-Org/ComfyUI/pull/17742): Request ID shown on polling timeouts
* [**Kling Avatar 2.0**](https://github.com/Comfy-Org/ComfyUI/pull/17740): Task ID shown on polling timeouts

### v0.45.3 (November 12, 2026)

**Partner Node Updates**

* [**Gemini 4 Hydrogen, Argon and Carbon**](https://github.com/Comfy-Org/ComfyUI/pull/17724): Three Gemini 4 models added to Google Gemini
* [**Seedance 2.5 4K**](https://github.com/Comfy-Org/ComfyUI/pull/17699): 4K no longer marked enterprise-only

### v0.45.2 (November 12, 2026)

**Bug Fixes**

* [**Workflow loading**](https://github.com/Comfy-Org/ComfyUI/pull/17675): Fixed some workflows not working properly


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.