v0.38.0

Comfy-Org/ComfyUIv0.38.0Sep 29, 2026by github-actions[bot]

Upgrading?

AI Summary

This release focuses on performance optimizations for models like Qwen and Flux, alongside significant improvements to asset management and database locking to prevent UI freezes. It also adds support for several new partner model nodes and features like HDR color space conversion.

Key Highlights

  • Optimized KV cache location and transformer blocks for Qwen 2.1
  • Enhanced asset scanning performance and database locking fixes
  • Added support for new partner models including GPT-6, Seedream, and Hunyuan Image 3.5

New Features

  • JsonExtractString now handles JSON with prefix and/or suffix
  • Added HDR LogC3 and ACEScct color space support
  • NPU async weight offload streams support
  • Removed torchaudio dependency
  • Added support for Ming image models and ID-V2V Wan 2.1

Full Release Notes

## What's Changed
* Improve qwen 2.1 KV cache location logic by @rattus128 in https://github.com/Comfy-Org/ComfyUI/pull/16429
* Compile Qwen Image 2.1 transformer blocks by @rattus128 in https://github.com/Comfy-Org/ComfyUI/pull/16430
* Allow model files to contain which attention should be used for each block. by @comfyanonymous in https://github.com/Comfy-Org/ComfyUI/pull/16419
* JsonExtractString can now handle JSON with prefix and/or suffix. by @comfyanonymous in https://github.com/Comfy-Org/ComfyUI/pull/16439
* Remove torchaudio dependency. by @comfyanonymous in https://github.com/Comfy-Org/ComfyUI/pull/16457
* Cuda graphs + memory compiler on ace step 1.5 autoregressive model. by @comfyanonymous in https://github.com/Comfy-Org/ComfyUI/pull/16461
* Blend H3 VAE tiles against composited neighbours by @kijai in https://github.com/Comfy-Org/ComfyUI/pull/16436
* Fast disk detection to all model loaders. by @comfyanonymous in https://github.com/Comfy-Org/ComfyUI/pull/16425
* [Partner Nodes] feat(Tencent): add Hunyuan Image 3.5 text-to-image and edit nodes by @comfyui-wiki in https://github.com/Comfy-Org/ComfyUI/pull/16462
* Update workflow templates to v0.11.68 by @comfyui-wiki in https://github.com/Comfy-Org/ComfyUI/pull/16466
* TextGenerate: Add system_prompt input and separate thinking output (CORE-460) by @kijai in https://github.com/Comfy-Org/ComfyUI/pull/16442
* MiniMax-H3: Support Fun-Controlnet-Union 2.0 (CORE-462) by @kijai in https://github.com/Comfy-Org/ComfyUI/pull/16471
* Fix MiniMax-H3 VAE rms_rope crash on offloaded qk_norm_scale by @chelsealong in https://github.com/Comfy-Org/ComfyUI/pull/16485
* Port some optimizations to flux model family. by @comfyanonymous in https://github.com/Comfy-Org/ComfyUI/pull/16488
* [Partner Nodes] feat(Quiver): add Arrow 2 models with reasoning effort widget by @bigcat88 in https://github.com/Comfy-Org/ComfyUI/pull/16478
* [Partner Nodes] feat(Anthropic): add Claude Opus 5.5 to the Claude node by @bigcat88 in https://github.com/Comfy-Org/ComfyUI/pull/16479
* [Partner Nodes] feat(Recraft): add Recraft V4.1 Flash to the V4 text-to-image node by @bigcat88 in https://github.com/Comfy-Org/ComfyUI/pull/16501
* Update workflow templates to v0.11.69 by @comfyui-wiki in https://github.com/Comfy-Org/ComfyUI/pull/16503
* Fix ImageUpscaleWithModel crashing on RGBA images by @chelsealong in https://github.com/Comfy-Org/ComfyUI/pull/16500
* Take the SQLite write lock up front for asset scan and output-registration writes by @synap5e in https://github.com/Comfy-Org/ComfyUI/pull/16480
* Move file reads out of the remaining asset write transactions by @synap5e in https://github.com/Comfy-Org/ComfyUI/pull/16486
* feat: ming-image support by @kijai in https://github.com/Comfy-Org/ComfyUI/pull/16482
* Fix saving/loading ming image model breaking detection. by @comfyanonymous in https://github.com/Comfy-Org/ComfyUI/pull/16514
* Lower memory usage and .comfy_attention support for lumina family models. by @comfyanonymous in https://github.com/Comfy-Org/ComfyUI/pull/16515
* [Partner Nodes] feat(OpenAI): add GPT-6 Sol and Luna models by @bigcat88 in https://github.com/Comfy-Org/ComfyUI/pull/16520
* [Partner Nodes] feat(ByteDance): add Seedream 5.0 Flash by @bigcat88 in https://github.com/Comfy-Org/ComfyUI/pull/16522
* fix(assets): harden the asset catalogue's failure paths by @synap5e in https://github.com/Comfy-Org/ComfyUI/pull/16393
* Add missing RDNA2 arches to list. by @comfyanonymous in https://github.com/Comfy-Org/ComfyUI/pull/16537
* feat(npu): support async weight offload streams by @Big2Wheel in https://github.com/Comfy-Org/ComfyUI/pull/16057
* Small fix: skip unnecessary work. by @comfyanonymous in https://github.com/Comfy-Org/ComfyUI/pull/16540
* Fix Qwen VL image preprocessing crash on 4-channel (RGBA) input by @chelsealong in https://github.com/Comfy-Org/ComfyUI/pull/16548
* [Partner Nodes] feat(ByteDance): add Seedance 2.5 Draft mode by @bigcat88 in https://github.com/Comfy-Org/ComfyUI/pull/16529
* Update workflow templates to v0.11.70 by @comfyui-wiki in https://github.com/Comfy-Org/ComfyUI/pull/16557
* Add HDR LogC3 and ACEScct to Convert Color Space node. by @comfyanonymous in https://github.com/Comfy-Org/ComfyUI/pull/16541
* Fix HDR output issue with negative values in Convert Image Color Space. by @comfyanonymous in https://github.com/Comfy-Org/ComfyUI/pull/16565
* Update sheetsage2 abc creation from upstream yue code. by @comfyanonymous in https://github.com/Comfy-Org/ComfyUI/pull/16569
* perf(assets): replace pathlib prefix checks in the asset scan by @synap5e in https://github.com/Comfy-Org/ComfyUI/pull/16543
* Mention some ComfyUI optimization details in README. by @comfyanonymous in https://github.com/Comfy-Org/ComfyUI/pull/16576
* Support tiny VAE for Qwen-Image 2.1 by @kijai in https://github.com/Comfy-Org/ComfyUI/pull/16552
* feat: Support ID-V2V Wan 2.1/VACE based model (CORE-426) by @kijai in https://github.com/Comfy-Org/ComfyUI/pull/15139
* Fix potential regression with previous PR. by @comfyanonymous in https://github.com/Comfy-Org/ComfyUI/pull/16596
* [Partner Nodes] chore(Sora): remove deprecated Sora nodes by @bigcat88 in https://github.com/Comfy-Org/ComfyUI/pull/16609
* Clamp Qwen Image 2.1 fp16 activations in place by @dcbert in https://github.com/Comfy-Org/ComfyUI/pull/16608
* Add comfy_attention and AttentionTensorContainer to a few models. by @comfyanonymous in https://github.com/Comfy-Org/ComfyUI/pull/16595
* Speedup Yue2 AR by @rattus128 in https://github.com/Comfy-Org/ComfyUI/pull/16626
* Searching "kitchen" now returns the attention node. by @comfyanonymous in https://github.com/Comfy-Org/ComfyUI/pull/16632
* Support Qwen-Image 2.1 union fun controlnet by @kijai in https://github.com/Comfy-Org/ComfyUI/pull/16519
* Update comfy-kitchen package version to 0.2.36 by @comfyanonymous in https://github.com/Comfy-Org/ComfyUI/pull/16635
* Optimize Seedvr2 by @kijai in https://github.com/Comfy-Org/ComfyUI/pull/16530
* Support w6a8 quantization format by @kijai in https://github.com/Comfy-Org/ComfyUI/pull/16483
* Fixes for last PR. by @comfyanonymous in https://github.com/Comfy-Org/ComfyUI/pull/16637
* fix(db): wait briefly for the database lock at startup by @synap5e in https://github.com/Comfy-Org/ComfyUI/pull/16602
* fix(assets): Start without the assets packages; --enable-assets reports what is missing and runs with assets disabled by @synap5e in https://github.com/Comfy-Org/ComfyUI/pull/16580
* perf(assets): keep the UI responsive during asset scans and rescan output by listing folders by @synap5e in https://github.com/Comfy-Org/ComfyUI/pull/16546
* Speedup generation on qwen3.5/3.8 with long contexts. by @comfyanonymous in https://github.com/Comfy-Org/ComfyUI/pull/16638
* Avoid unnecessary write. by @comfyanonymous in https://github.com/Comfy-Org/ComfyUI/pull/16639
* Update codeowners. by @comfyanonymous in https://github.com/Comfy-Org/ComfyUI/pull/16655

## New Contributors
* @Big2Wheel made their first contribution in https://github.com/Comfy-Org/ComfyUI/pull/16057
* @dcbert made their first contribution in https://github.com/Comfy-Org/ComfyUI/pull/16608

**Full Changelog**: https://github.com/Comfy-Org/ComfyUI/compare/v0.37.0...v0.38.0

Related Repositories