v0.9.2

What's Changed

• [API Nodes] Gemini: raise exception when no candidates due to safety block by @bigcat88 in #11848
• Support "lite" version of alibaba-pai Z-Image Controlnet by @Acly in #11849
• Load metadata on VAELoader by @kijai in #11846
• add blueprints dir for built-in blueprints by @ric-yu in #11853
• [Api Nodes]: Improve Price Badge Declarations by @bigcat88 in #11582
• Optimize nvfp4 lora applying. by @comfyanonymous in #11854
• Optimize nvfp4 lora applying. by @comfyanonymous in #11856
• Adds crop to multiple mode to ResizeImageMaskNode. by @nomadoor in #11838
• fix: Update repository reference comfyanonymous/ComfyUI → Comfy-Org/ComfyUI by @christian-byrne in #11858
• Fix scale_shorter_dimension for portrait inputs by @nomadoor in #11862
• feat: add CI container version bump automation by @Myestery in #11692
• feat: throttle ProgressBar updates to reduce WebSocket flooding by @silveroxides in #11504
• Optimize nvfp4 lora applying. by @comfyanonymous in #11866
• Fix VAELoader by @comfyanonymous in #11880
• feat(api-nodes): add Meshy 3D nodes by @bigcat88 in #11843
• utils: fix lanczos grayscale upscaling by @rattus128 in #11873
• Reduce log spam when loading LTX2 embeddings_connector weights by @kijai in #11874
• Flux2 Klein support. by @comfyanonymous in #11890

New Contributors

• @nomadoor made their first contribution in #11838

Full Changelog: v0.9.1...v0.9.2

v0.9.1

What's Changed

• Bump ltxav mem estimation a bit. by @comfyanonymous in #11842

Full Changelog: v0.9.0...v0.9.1

v0.9.0

What's Changed

• Revert "Force sequential execution in CI test jobs" by @yoland68 in #11725
• bump comfyui_manager version to the 4.0.5 by @ltdrdata in #11732
• Better detection if AMD torch compiled with efficient attention. by @comfyanonymous in #11745
• Fix import issue. by @comfyanonymous in #11746
• Add node: JoinAudioChannels by @kijai in #11728
• Fix csp error in frontend when forcing offline. by @comfyanonymous in #11749
• Bump comfyui-frontend-package to 1.36.13 by @comfy-pr-bot in #11645
• Add most basic Asset support for models by @Kosinkadink in #11315
• Fix VAEEncodeForInpaint to support WAN VAE tuple downscale_ratio by @rattus128 in #11572
• add node - image compare by @jtydhr88 in #11343
• feat: add 'cancelled' status for interrupted jobs in /api/jobs by @ric-yu in #11680
• Add workaround for a hacky nodepack existing by @Kosinkadink in #11755
• Be less strict when loading mixed ops weights. by @comfyanonymous in #11769
• fix(api-nodes): do not downscale the input image for Topaz Enhance by @bigcat88 in #11768
• [API Nodes] add nodes for Vidu2 by @bigcat88 in #11760
• Update embedded docs to v0.4.0 by @comfyui-wiki in #11776
• Properly save mixed ops. by @comfyanonymous in #11772
• Enable PyTorch Attention by default on gfx1200 by @0xDELUXA in #11793
• Fix issue with t5 text encoder in fp4. by @comfyanonymous in #11794
• Fix chroma fp8 te being treated as fp16. by @comfyanonymous in #11795
• fix(api-nodes): use a unique name for uploading audio files by @bigcat88 in #11778
• Put more details about portable in readme. by @comfyanonymous in #11816
• Support ModelScope-Trainer/DiffSynth LoRA format for Z-Image models by @kelseyee in #11805
• Support the siglip 2 naflex model as a clip vision model. by @comfyanonymous in #11831
• Reduce LTX2 VRAM use by more efficient timestep embed handling by @kijai in #11829
• Update workflow templates to v0.8.0 by @comfyui-wiki in #11828
• Update workflow templates to v0.8.4 by @comfyui-wiki in #11835
• Make loras work on nvfp4 models. by @comfyanonymous in #11837
• Refactor to try to lower mem usage. by @comfyanonymous in #11840
• Make database bulk_ops.py more widely compatible across sqlalchemy+python versions by @Kosinkadink in #11839
• Patch frontend to 1.36.14 (from 1.36.13) by @christian-byrne in #11841

New Contributors

• @kelseyee made their first contribution in #11805

Full Changelog: v0.8.2...v0.9.0

v0.8.2

What's Changed

• Tweak ltxv vae mem estimation. by @comfyanonymous in #11722

Full Changelog: v0.8.1...v0.8.2

v0.8.1

What's Changed

• Fix stable release workflow not pulling latest comfy kitchen. by @comfyanonymous in #11695
• Support gemma 12B with quant weights. by @comfyanonymous in #11696
• model_patcher: Remove confusing load stat by @rattus128 in #11710
• Add device selection for LTXAVTextEncoderLoader by @kijai in #11700
• Lower ltxv text encoder vram use. by @comfyanonymous in #11713
• Bump required comfy-kitchen version. by @comfyanonymous in #11714
• Increase ltxav mem estimation by a bit. by @comfyanonymous in #11715
• Add memory estimation function to ltxav text encoder. by @comfyanonymous in #11716
• ops: Fix offloading with FP8MM performance by @rattus128 in #11697
• Add warning for old pytorch and comfy kitchen. Update your pytorch to cu130 to get a performance boost. by @comfyanonymous in #11718
• Update template to 0.7.69 by @comfyui-wiki in #11719

Full Changelog: v0.8.0...v0.8.1

v0.8.0

What's Changed

• Add support for sage attention 3 in comfyui, enable via new cli arg by @mengqin in #11026
• V3 Improvements + DynamicCombo + Autogrow exposed in public API by @Kosinkadink in #11345
• fix(api-nodes-vidu): preserve percent-encoding for signed URLs by @bigcat88 in #11564
• Update workflow templates to v0.7.65 by @comfyui-wiki in #11579
• Refactor: move clip_preprocess to comfy.clip_model by @comfyanonymous in #11586
• Remove duplicate import of model_management by @comfyanonymous in #11587
• New Year ruff cleanup. by @comfyanonymous in #11595
• Ignore all frames except the first one for MPO format. by @bigcat88 in #11569
• Give Mahiro CFG a more appropriate display name by @throttlekitty in #11580
• [API Nodes] Tripo3D: pass face_limit parameter only when it differs from default by @bigcat88 in #11601
• Remove leftover scaled_fp8 key. by @comfyanonymous in #11603
• Print memory summary on OOM to help with debugging. by @comfyanonymous in #11613
• [API Nodes] add support for 720p resolution for Kling Omni nodes by @bigcat88 in #11604
• Fix case where upscale model wouldn't be moved to cpu. by @comfyanonymous in #11633
• Support the LTXV 2 model. by @comfyanonymous in #11632
• Add LTXAVTextEncoderLoader node. by @comfyanonymous in #11634
• Refactor module_size function. by @comfyanonymous in #11637
• Fix name. by @comfyanonymous in #11638
• Initial ops changes to use comfy_kitchen: Initial nvfp4 checkpoint support. by @comfyanonymous in #11635
• Use rope functions from comfy kitchen. by @comfyanonymous in #11647
• Revert "Use rope functions from comfy kitchen." by @comfyanonymous in #11648
• Update workflow templates to v0.7.66 by @comfyui-wiki in #11652
• Add helpful message to portable. by @comfyanonymous in #11671
• Update comfy-kitchen version to 0.2.1 by @comfyanonymous in #11672
• Use rope functions from comfy kitchen. by @comfyanonymous in #11674
• Update workflow templates to v0.7.67 by @comfyui-wiki in #11667
• Fix lowvram issue with ltxv2 text encoder. by @comfyanonymous in #11675
• Disable ltxav previews. by @comfyanonymous in #11676
• Skip fp4 matrix mult on devices that don't support it. by @comfyanonymous in #11677
• Disable comfy kitchen cuda if pytorch cuda less than 13 by @comfyanonymous in #11681
• Update comfy-kitchen version to 0.2.3 by @comfyanonymous in #11685
• Force sequential execution in CI test jobs by @yoland68 in #11687
• [API Nodes] add WAN2.6 ReferenceToVideo by @bigcat88 in #11644
• Fix fp8 fast issue. by @comfyanonymous in #11688

New Contributors

• @mengqin made their first contribution in #11026
• @throttlekitty made their first contribution in #11580

Full Changelog: v0.7.0...v0.8.0

v0.7.0

What's Changed

• Bump comfyui-frontend-package to 1.35.9 by @comfy-pr-bot in #11470
• Update workflow templates to v0.7.64 by @comfyui-wiki in #11496
• Add a ManualSigmas node. by @comfyanonymous in #11499
• Specify in readme that we only support pytorch 2.4 and up. by @comfyanonymous in #11512
• bump comfyui_manager version to the 4.0.4 by @ltdrdata in #11521
• Fix noise with ancestral samplers when inferencing on cpu. by @comfyanonymous in #11528
• feat(api-nodes): add Kling Motion Control node by @bigcat88 in #11493
• [V3] converted nodes_images.py to V3 schema by @bigcat88 in #11206
• fix(api-nodes-gemini): always force enhance_prompt to be True by @bigcat88 in #11503
• chore(api-nodes): move to credits from $ by @bigcat88 in #11489
• Enable async offload by default for AMD. by @comfyanonymous in #11534
• Comment out unused norm_final in lumina/z image model. by @comfyanonymous in #11545
• mm: discard async errors from pinning failures by @rattus128 in #10738
• Add some warnings for pin and unpin errors. by @comfyanonymous in #11561
• ResizeByLongerSide: support video by @tavihalperin in #11555
• Mark "seededit" as deprecated, adjust display name of Seedream node by @bigcat88 in #11490
• Add handling for vace_context in context windows by @drozbay in #11386

New Contributors

• @tavihalperin made their first contribution in #11555

Full Changelog: v0.6.0...v0.7.0

v0.6.0

What's Changed

• Resolution bucketing and Trainer implementation refactoring by @KohakuBlueleaf in #11117
• skip_load_model -> force_full_load by @comfyanonymous in #11390
• Add unified jobs API with /api/jobs endpoints by @ric-yu in #11054
• Update workflow templates to v0.7.60 by @comfyui-wiki in #11403
• Support loading Wan/Qwen VAEs with different in/out channels. by @comfyanonymous in #11405
• Trim/pad channels in VAE code. by @comfyanonymous in #11406
• Diffusion model part of Qwen Image Layered. by @comfyanonymous in #11408
• Add LatentCutToBatch node. by @comfyanonymous in #11411
• [API Nodes] add Flux2MaxImage API Node by @bigcat88 in #11420
• bump comfyui_manager version to the 4.0.3b7 by @ltdrdata in #11422
• Allow enabling use of MIOpen by setting an env var by @BradPepersAMD in #11366
• Support nested tensor denoise masks. by @comfyanonymous in #11431
• Fix error from logging line in Context Windows by @drozbay in #11423
• Implement sliding attention in Gemma3 by @woct0rdho in #11409
• Disable prompt weights on newbie te. by @comfyanonymous in #11434
• ZImageFunControlNet: Fix mask concatenation in --gpu-only by @rattus128 in #11421
• Fix issue with batches and newbie. by @comfyanonymous in #11435
• Implement Jina CLIP v2 and NewBie dual CLIP by @woct0rdho in #11415
• Only apply gemma quant config to gemma model for newbie. by @comfyanonymous in #11436
• chore(api-nodes): by default set Watermark generation to False by @bigcat88 in #11437
• fix(api-nodes): Topaz 4k video upscaling by @bigcat88 in #11438
• Core release process. by @comfyanonymous in #11447
• Add node to create empty latents for qwen image layered model. by @comfyanonymous in #11460
• [API Nodes] extend possible duration range for Kling O1 StartEndFrame node by @bigcat88 in #11451
• Update workflow templates to v0.7.62 by @comfyui-wiki in #11467
• Make denoised output on custom sampler nodes work with nested tensors. by @comfyanonymous in #11471
• api-nodes: use new custom endpoint for Nano Banana by @bigcat88 in #11311
• Update workflow templates to v0.7.63 by @comfyui-wiki in #11482

New Contributors

• @BradPepersAMD made their first contribution in #11366
• @woct0rdho made their first contribution in #11409

Full Changelog: v0.5.1...v0.6.0

v0.5.1

What's Changed

• [API Nodes] GPT-Image-1.5 by @bigcat88 in #11368
• fix regression in V3 nodes processing by @bigcat88 in #11375
• SA-Solver: return the result of the last predictor step correctly by @chaObserv in #11380
• Better handle torch being imported by prestartup nodes. by @comfyanonymous in #11383

Full Changelog: v0.5.0...v0.5.1

v0.5.0

What's Changed

• Bump frontend to 1.34.8 by @benceruleanlu in #11220
• fix: correctly process the Node V1 dict results on V3 node by @bigcat88 in #11237
• Tweak Z Image memory estimation. by @comfyanonymous in #11254
• Fix: filter hidden files from /internal/files endpoint by @Myestery in #11191
• Lower VAE loading requirements：Create a new branch for GPU memory calculations in qwen-image vae by @jiangchengchengark in #11199
• feat(api-nodes): enable Kling Omni Image node by @bigcat88 in #11229
• Adjust memory usage factor for sd3. by @comfyanonymous in #11257
• Fix regular chroma radiance by @comfyanonymous in #11276
• This only works on radiance. by @comfyanonymous in #11277
• Better chroma radiance vram estimation. by @comfyanonymous in #11278
• Make portable updater work with repos in unmerged state. by @comfyanonymous in #11281
• WanMove support by @kijai in #11247
• Respect the dtype the op was initialized in for non quant mixed op. by @comfyanonymous in #11282
• feat(api-nodes): Kling TextToVideoWithAudio and ImageToVideoWithAudio nodes by @bigcat88 in #11267
• Fix bias dtype issue in mixed ops. by @comfyanonymous in #11293
• Basic implementation of z image fun control union 2.0 by @comfyanonymous in #11304
• Fix pytorch warnings. by @comfyanonymous in #11314
• seeds_2: add phi_2 variant and sampler node by @chaObserv in #11309
• Update warning for old pytorch version. by @comfyanonymous in #11319
• bump manager to the 4.0.3b5 by @ltdrdata in #11324
• Update workflow templates to v0.7.59 by @comfyui-wiki in #11337
• Disable guards on transformer_options when torch.compile by @comfyanonymous in #11317
• [API Nodes] drop very old Kling models by @bigcat88 in #11307
• drop Pika API nodes by @bigcat88 in #11306
• feat(preview): add per-queue live preview method override by @ltdrdata in #11261
• comfy_api: remove usage of "Type","List" and "Dict" types by @bigcat88 in #11238
• Add context windows callback for custom cond handling by @drozbay in #11208
• Support the new qwen edit 2511 reference method. by @comfyanonymous in #11340
• [add] tripo3.0 by @seed93 in #10663
• Add code to detect if a z image fun controlnet is broken or not. by @comfyanonymous in #11341
• [BlockInfo] Lumina by @Haoming02 in #11227
• [BlockInfo] Wan by @Haoming02 in #10845
• Only enable fp16 on ZImage on newer pytorch. by @comfyanonymous in #11344
• Patch frontend to 1.34.9 (from 1.34.8) by @christian-byrne in #11342
• Inpainting for z image fun control. Use the ZImageFunControlnet node. by @comfyanonymous in #11346
• Add a way to set the default ref method in the qwen image code. by @comfyanonymous in #11349
• Update workflows for new release process by @benceruleanlu in #11064
• [API Nodes] add Wan2.6 model to video nodes by @bigcat88 in #11357
• Check state dict key to auto enable the index_timestep_zero ref method. by @comfyanonymous in #11362
• Add exp_heun_2_x0 sampler series by @chaObserv in #11360

New Contributors

• @jiangchengchengark made their first contribution in #11199
• @drozbay made their first contribution in #11208
• @seed93 made their first contribution in #10663

Full Changelog: v0.4.0...v0.5.0