What's Changed
- Add AGENTS.md by @comfyanonymous in #14696
- Add some more stuff to AGENTS.md by @comfyanonymous in #14704
- Fix Qwen3-VL tokenizer crash with custom embeddings by @silveroxides in #14713
- Update AGENTS.md with more stuff. by @comfyanonymous in #14725
- [Partner Nodes] chore(Ideogram): remove IdeogramV1 and IdeogramV2 nodes by @bigcat88 in #14712
- Update AGENTS.md by @comfyanonymous in #14726
- security: fix four vulnerabilities (GHSA-779p-m5rp-r4h4) by @mattmillerai in #14734
- Update AGENTS.md by @comfyanonymous in #14733
- [Partner Nodes] feat(ByteDance): add support for Seed Audio 1.0 by @bigcat88 in #14731
- Update workflow templates to v0.11.2 by @comfyui-wiki in #14741
- [Partner Nodes] chore(StabilityAI): remove StabilityAI nodes by @bigcat88 in #14737
- Update AGENTS.md by @comfyanonymous in #14738
- [Partner Nodes] chore(ByteDance): adjust category name by @bigcat88 in #14752
- Add CLAUDE.md as symlink to AGENTS.md by @robinjhuang in #14757
- fix: error when embedding is loaded with models using llama_template by @silveroxides in #14744
- Try to make coderabbit enforce AGENTS.md by @comfyanonymous in #14759
- fix: Bug when setting transparency in color picker by @alexisrolland in #14764
- [Partner Nodes] fix(logs-auth): mask authorization headers in logs by @bigcat88 in #14774
- Small speedup for text model sampling. by @comfyanonymous in #14773
- Update embedded docs to v0.5.7 by @comfyui-wiki in #14783
- Skip unloading dynamic model patchers in current workflow. by @comfyanonymous in #14799
- ci: add CLA Assistant workflow by @alexisrolland in #14582
- Delete symlink that breaks our updates. by @comfyanonymous in #14803
- fix(Video): don't crash on videos with undecodable audio streams by @bigcat88 in #14746
- feat: Implement basic text overlay node (CORE-137) by @drozbay in #14610
- Update AGENTS.md by @comfyanonymous in #14819
- Support gqa on all attention backends, drop support for pytorch 2.4 by @comfyanonymous in #14772
- Fix crash when using UNetSelfAttentionMultiply by @comfyanonymous in #14823
- add models-directory launch argument by @silveroxides in #9113
- Fix AttributeError in VAE.is_dynamic() for VAEs constructed without a patcher by @j2gg0s in #14826
- [Partner Nodes] feat(ByteDance): add Seedream 5 Pro model support by @bigcat88 in #14832
- Update workflow templates to v0.11.6 by @comfyui-wiki in #14834
- feat(assets): add namespaced model_type tags and align tag semantics by @synap5e in #14511
- Fix qwen3vl reference images when used as a text encode models. by @comfyanonymous in #14845
- feat: allow --comfy-api-base target ephemeral testenvs by @bigcat88 in #14569
- fix: Load3D failing path validation from double path resolution by @jtydhr88 in #14852
- feat(api): expose effective extension filter on /experiment/models by @synap5e in #14797
- Update cla.yml by @alexisrolland in #14851
- Support convrot int4 models. by @comfyanonymous in #14859
- Try to fix the model reloading issue some people have. by @comfyanonymous in #14822
- Fix FP8 activation quantization for >2D activations in mixed_precision_ops by @liminfei-amd in #14643
- Enable comfy-kitchen Triton backend by default on ROCm/AMD by @liminfei-amd in #14862
- Fix black image on turing when using int4 models. by @comfyanonymous in #14864
- Add SeedVR2 support (CORE-6) by @pollockjj in #14424
- [Partner Nodes] feat(Seedream): add widget to disable thinking by @bigcat88 in #14853
- [Partner Nodes] fix(Tencent): restore Tencent3DPartNode FBX output by @bigcat88 in #14867
- Only auto-enable the ROCm comfy-kitchen Triton backend on matrix-core GPUs by @liminfei-amd in #14869
- Save Text Node (CORE-176) by @yousef-rafat in #14102
- feat: add bboxes input to Create Bounding Boxes node by @jtydhr88 in #14724
- CORE-329 feat: add Save 3D (Advanced) node family by @jtydhr88 in #14701
- fix: Undeprecate the text node by @alexisrolland in #14870
- chore: Update preview nodes by @alexisrolland in #14871
- Try to fix some issues with the seedvr VAE. by @comfyanonymous in #14877
- Temporarily disable auto enabling triton by default on AMD. by @comfyanonymous in #14878
- Try to fix flash attention related issue on AMD. by @comfyanonymous in #14880
- Fix SaveVideo description: says images, saves video by @gmhelmold in #14885
- Support PID 1.5 models. by @comfyanonymous in #14894
- [Partner Nodes] feat(client): send ComfyUI Core version in request headers by @bigcat88 in #14910
- fix(image): correct HLG inverse-OETF clamp in hlg_to_linear by @bigcat88 in #14762
- fix(image): support single-channel images in Save Image (Advanced) by @bigcat88 in #14761
- [Partner Nodes] chore(Google): reroute Gemini Image preview models to release versions by @bigcat88 in #14917
- Update embedded docs to v0.5.8 by @comfyui-wiki in #14920
- chore(openapi): sync shared API contract from cloud@491db61 by @comfy-pr-bot in #14815
- Update workflow templates to v0.11.9 by @comfyui-wiki in #14924
- Fix hidream o1 regression. by @comfyanonymous in #14923
- More optimized int8 and int4 on turing. by @comfyanonymous in #14927
- Fix 3D advanced nodes crashing in API mode (no UI) by @jtydhr88 in #14930
- Fix int8 performance regression on 16xx series. by @comfyanonymous in #14941
- [Partner Nodes] feat(sync.so): add support for "sync-3" model by @bigcat88 in #14928
- Fix cached outputs missing from job results when prompt has no client_id by @bigcat88 in #14939
- Bump comfyui-frontend-package to 1.45.21 by @comfy-pr-bot in #14944
New Contributors
- @j2gg0s made their first contribution in #14826
- @liminfei-amd made their first contribution in #14643
- @gmhelmold made their first contribution in #14885
Full Changelog: v0.27.0...v0.28.0