v0.23.0

google/magikav0.23.0Jun 1, 2026by github-actions[bot]

AI Summary

Significant performance improvements include multi-threaded model loading and offloading to disk, alongside new integrations for MediaPipe face detection, Microsoft Lens, and PiD. API security and OpenAPI updates are also included.

Key Highlights

  • Multi-threaded model loading and disk offloading for speedups
  • MediaPipe face detection integration
  • Support for NVIDIA PixelDiT (PiD) and Microsoft Lens
  • OpenAPI OAuth 2.1 endpoints
  • MultiGPU Work Units for accelerated sampling

New Features

  • MediaPipe face detection
  • Multi-threaded model loading
  • PiD and Lens support
  • OpenAPI OAuth 2.1 + DCR endpoints
  • MultiGPU Work Units
  • Logic nodes (And, Or, Not)
  • Save Image advanced node
  • Colored logs

Full Release Notes

## What's Changed
* feat: MediaPipe face detection (CORE-235) by @kijai in https://github.com/Comfy-Org/ComfyUI/pull/14009
* Multi-threaded load of models from disk (big load time speedups & Offload to disk) (CORE-43,CORE-152,CORE-164,CORE-165,CORE-117) by @rattus128 in https://github.com/Comfy-Org/ComfyUI/pull/13802
* Repo security stuff. by @comfyanonymous in https://github.com/Comfy-Org/ComfyUI/pull/14019
* feat(openapi): add OAuth 2.1 + RFC 7591 DCR endpoints by @mattmillerai in https://github.com/Comfy-Org/ComfyUI/pull/14026
* fix(openapi): correct POST /api/assets/import to importPublishedAssets by @mattmillerai in https://github.com/Comfy-Org/ComfyUI/pull/14027
* openapi: remove cloud-only job_ids query param from GET /api/assets by @mattmillerai in https://github.com/Comfy-Org/ComfyUI/pull/14016
* Update MediaPipe nodes to standardize with existing code base (CORE-242) by @alexisrolland in https://github.com/Comfy-Org/ComfyUI/pull/14025
* chore: Dataset nodes clean-up (CORE-237) by @alexisrolland in https://github.com/Comfy-Org/ComfyUI/pull/14002
* Update MoGe nodes display names, search aliases and descriptions by @alexisrolland in https://github.com/Comfy-Org/ComfyUI/pull/14030
* fix(lanczos): correct dimension transposition for single-channel tensors (CORE-228) by @edoardocarmignani in https://github.com/Comfy-Org/ComfyUI/pull/12679
* Fix reshaping lora application by @rattus128 in https://github.com/Comfy-Org/ComfyUI/pull/14031
* [Partner Nodes] add OpenRouter LLM node by @bigcat88 in https://github.com/Comfy-Org/ComfyUI/pull/14007
* Update workflow templates to v0.9.82 by @comfyui-wiki in https://github.com/Comfy-Org/ComfyUI/pull/14034
* [Partner Nodes] add widget for auto upscaling for the ByteDance2Reference node by @bigcat88 in https://github.com/Comfy-Org/ComfyUI/pull/14032
* Add backport release workflow. by @comfyanonymous in https://github.com/Comfy-Org/ComfyUI/pull/14038
* Fix SIGPIPE false negative in Backport Release workflow validation by @Kosinkadink in https://github.com/Comfy-Org/ComfyUI/pull/14041
* Add validation for source branch in backport workflow by @comfyanonymous in https://github.com/Comfy-Org/ComfyUI/pull/14042
* Update backport workflow to use commit SHA input by @comfyanonymous in https://github.com/Comfy-Org/ComfyUI/pull/14043
* fe: Consolidate warnings spam by @rattus128 in https://github.com/Comfy-Org/ComfyUI/pull/13970
* Update Discord invite link in README.md by @comfyanonymous in https://github.com/Comfy-Org/ComfyUI/pull/14045
* Adding in And, Or, and Not nodes. by @Pauan in https://github.com/Comfy-Org/ComfyUI/pull/14004
* Move logic nodes into utils category by @alexisrolland in https://github.com/Comfy-Org/ComfyUI/pull/14033
* [Partner Nodes] add new Rodin2.5 nodes by @bigcat88 in https://github.com/Comfy-Org/ComfyUI/pull/14051
* openapi: align response declarations with implementation (5 endpoints) by @mattmillerai in https://github.com/Comfy-Org/ComfyUI/pull/14058
* Delete the source branch after doing the backport. by @comfyanonymous in https://github.com/Comfy-Org/ComfyUI/pull/14062
* openapi: rename 55 cloud-side operationIds to match runtime (PR A of 3) by @mattmillerai in https://github.com/Comfy-Org/ComfyUI/pull/14060
* openapi: rename cloud-side response schemas to match runtime (PR D) by @mattmillerai in https://github.com/Comfy-Org/ComfyUI/pull/14065
* openapi: add enum values + FeedbackRequest schema for cloud cutover (PR E) by @mattmillerai in https://github.com/Comfy-Org/ComfyUI/pull/14070
* Remove dead code. by @comfyanonymous in https://github.com/Comfy-Org/ComfyUI/pull/14072
* Bump comfyui-frontend-package to 1.44.19 by @comfy-pr-bot in https://github.com/Comfy-Org/ComfyUI/pull/14074
* Fix --use-flash-attention ignored when xformers installed. by @comfyanonymous in https://github.com/Comfy-Org/ComfyUI/pull/14083
* fix: correct description of where compiled FE files live by @robinjhuang in https://github.com/Comfy-Org/ComfyUI/pull/14013
* openapi: fix invalid BillingStatus schema (object + enum hybrid) by @mattmillerai in https://github.com/Comfy-Org/ComfyUI/pull/14071
* Aimdo 0.4.5 + Cold marking of pins by @rattus128 in https://github.com/Comfy-Org/ComfyUI/pull/14088
* cache-ram: lower thresholds by @rattus128 in https://github.com/Comfy-Org/ComfyUI/pull/14089
* Add colored logs by @Talmaj in https://github.com/Comfy-Org/ComfyUI/pull/14036
* Save Image advanced node (CORE-32) by @comfyanonymous in https://github.com/Comfy-Org/ComfyUI/pull/13850
* Update embedded docs to v0.5.1 by @comfyui-wiki in https://github.com/Comfy-Org/ComfyUI/pull/14101
* Emit `hash` alongside `asset_hash` on all Asset responses by @mattmillerai in https://github.com/Comfy-Org/ComfyUI/pull/13739
* Add new open-source model and built-in tool blueprints by @comfyui-wiki in https://github.com/Comfy-Org/ComfyUI/pull/13980
* MultiGPU Work Units For Accelerated Sampling (CORE-184) by @Kosinkadink in https://github.com/Comfy-Org/ComfyUI/pull/7063
* Remove some useless annotations imports. by @comfyanonymous in https://github.com/Comfy-Org/ComfyUI/pull/14105
* multigpu: use unet_manual_cast for SelectModelDevice compute dtype by @Kosinkadink in https://github.com/Comfy-Org/ComfyUI/pull/14108
* fix(LTX AV): pass correct timesteps to cross-attention AdaLN modulation by @izorinLightricks in https://github.com/Comfy-Org/ComfyUI/pull/14097
* feat: Microsoft Lens support (CORE-248) by @kijai in https://github.com/Comfy-Org/ComfyUI/pull/14077
* Lens: some cleanup by @kijai in https://github.com/Comfy-Org/ComfyUI/pull/14112
* openapi: document QueueManageResponse body on POST /api/queue by @mattmillerai in https://github.com/Comfy-Org/ComfyUI/pull/14117
* feat: Support NVIDIA PixelDiT and PiD (CORE-201) by @kijai in https://github.com/Comfy-Org/ComfyUI/pull/14103
* Closer memory usage factors for PID by @comfyanonymous in https://github.com/Comfy-Org/ComfyUI/pull/14123
* Add memory usage factor for lens model. by @comfyanonymous in https://github.com/Comfy-Org/ComfyUI/pull/14124
* fix: Stop LTXVCropGuides leaving stray latent frames when guides share a start position (CORE-206) by @drozbay in https://github.com/Comfy-Org/ComfyUI/pull/13882
* [Partner Nodes] feat: improve video references uploading for SeeDance 2 by @bigcat88 in https://github.com/Comfy-Org/ComfyUI/pull/14098
* Update workflow templates to v0.9.85 by @comfyui-wiki in https://github.com/Comfy-Org/ComfyUI/pull/14134
* [Partner Nodes] feat: add Krea2 Image nodes by @bigcat88 in https://github.com/Comfy-Org/ComfyUI/pull/14130
* [V3] convert nodes_lt_upsampler nodes to V3 schema by @bigcat88 in https://github.com/Comfy-Org/ComfyUI/pull/12423
* [Partner Nodes] feat: Beeble SwitchX nodes by @bigcat88 in https://github.com/Comfy-Org/ComfyUI/pull/14137
* Support context window for PiD and fix lq_latent rounding by @kijai in https://github.com/Comfy-Org/ComfyUI/pull/14136
* Disable sage attention in stable audio dit and VAE. by @comfyanonymous in https://github.com/Comfy-Org/ComfyUI/pull/14148
* chore: Update nodes categories by @alexisrolland in https://github.com/Comfy-Org/ComfyUI/pull/14145
* feat: add camera intrinsics fields to Load3DCamera info by @jtydhr88 in https://github.com/Comfy-Org/ComfyUI/pull/14143
* Make Load3D model_file optional by adding "none" choice by @jtydhr88 in https://github.com/Comfy-Org/ComfyUI/pull/13379
* [Partner Nodes] fix: add runtime check for SeeDance2 image inputs by @bigcat88 in https://github.com/Comfy-Org/ComfyUI/pull/14152
* [Partner Nodes] chore: update the category of the Beeble nodes by @bigcat88 in https://github.com/Comfy-Org/ComfyUI/pull/14156
* feat: add overwrite/increment to SaveImageTextDataSetToFolderNode by @zeerd in https://github.com/Comfy-Org/ComfyUI/pull/13215
* Add unreviewed merge detector for SOC 2 compliance by @luke-mino-altherr in https://github.com/Comfy-Org/ComfyUI/pull/14146
* Bump pyav package to fix some image loading issues. by @comfyanonymous in https://github.com/Comfy-Org/ComfyUI/pull/14160
* refactor: drop rotation from Load3DCamera by @jtydhr88 in https://github.com/Comfy-Org/ComfyUI/pull/14159
* Update and pin comfy-kitchen version to 0.2.9 by @comfyanonymous in https://github.com/Comfy-Org/ComfyUI/pull/14161
* float: use CK stochastic rounding cuda kernel (CORE-200) by @rattus128 in https://github.com/Comfy-Org/ComfyUI/pull/13971
* Update workflow templates to v0.9.91 by @comfyui-wiki in https://github.com/Comfy-Org/ComfyUI/pull/14163
* feat: add model_info output to Load3D node by @jtydhr88 in https://github.com/Comfy-Org/ComfyUI/pull/14144
* [Partner Nodes] fix: removed "beta" models versions from Grok nodes by @bigcat88 in https://github.com/Comfy-Org/ComfyUI/pull/14170
* Fix background removal mask output shape by @kijai in https://github.com/Comfy-Org/ComfyUI/pull/14171
* [Partner Nodes] feat: add new nodes for Tripo3D P1 model by @bigcat88 in https://github.com/Comfy-Org/ComfyUI/pull/14155
* Remove useless code. by @comfyanonymous in https://github.com/Comfy-Org/ComfyUI/pull/14178
* Speed up anima a bit on nvidia. by @comfyanonymous in https://github.com/Comfy-Org/ComfyUI/pull/14181
* feat(io): add File3DPLY / File3DSPLAT / File3DSPZ / File3DKSPLAT types by @jtydhr88 in https://github.com/Comfy-Org/ComfyUI/pull/14185
* Threaded Loader performance fixes / improvements (+ Aimdo 0.4.6) by @rattus128 in https://github.com/Comfy-Org/ComfyUI/pull/14116
* Revert deprecation of non-dynamic smart memory (CORE-152 (revert)) by @rattus128 in https://github.com/Comfy-Org/ComfyUI/pull/14183
* feat: add Preview3DAdvanced node by @jtydhr88 in https://github.com/Comfy-Org/ComfyUI/pull/14175
* Update embedded docs to v0.5.2 by @comfyui-wiki in https://github.com/Comfy-Org/ComfyUI/pull/14193
* Speed up ernie model by a bit on nvidia and use higher quality rope. by @comfyanonymous in https://github.com/Comfy-Org/ComfyUI/pull/14192
* fix: enable XPU MultiGPU CFG Split on Windows (CFG > 1) by @savvadesogle in https://github.com/Comfy-Org/ComfyUI/pull/14191
* feat: Add gaussian splat nodes by @kijai in https://github.com/Comfy-Org/ComfyUI/pull/14190
* [Partner Nodes] added grok-imagine-video-1.5 model to the GrokVideo node in First Frame mode by @bigcat88 in https://github.com/Comfy-Org/ComfyUI/pull/14198
* Remove old portable updater migration code. by @comfyanonymous in https://github.com/Comfy-Org/ComfyUI/pull/14202
* feat: Add TripoSplat support by @kijai in https://github.com/Comfy-Org/ComfyUI/pull/14210
* [Partner Nodes] feat: add Flux Virtual Try-On and Erase nodes by @bigcat88 in https://github.com/Comfy-Org/ComfyUI/pull/14207
* Update workflow templates to v0.9.92 by @comfyui-wiki in https://github.com/Comfy-Org/ComfyUI/pull/14212
* [Partner Nodes] fix: respect VideoSlice trim when resizing videos by @bigcat88 in https://github.com/Comfy-Org/ComfyUI/pull/14213

## New Contributors
* @edoardocarmignani made their first contribution in https://github.com/Comfy-Org/ComfyUI/pull/12679
* @zeerd made their first contribution in https://github.com/Comfy-Org/ComfyUI/pull/13215
* @savvadesogle made their first contribution in https://github.com/Comfy-Org/ComfyUI/pull/14191

**Full Changelog**: https://github.com/Comfy-Org/ComfyUI/compare/v0.22.0...v0.23.0