ComfyUI May Wrapped: A Look at the New Models and the Growing ComfyHub Ecosystem
ComfyUI has just released its May wrap‑up, detailing eleven new models that span video, image, 3D, audio, and multimodal capabilities. The update also highlights the ongoing expansion of ComfyHub, the community-driven workflow gallery that is beginning to collect more than 500 ready‑to‑use templates.
What ComfyUI Is Announcing
In the official ComfyUI Blog post (June 1 2026), the team enumerates a suite of freshly integrated models, each bringing unique features to the workflow builder. The models are available as Partner Nodes or built‑in functionality, ready for immediate use in any ComfyUI graph.
Key Features in This Release
| Model | Provider | Core Capability | Notable Highlights |
|---|---|---|---|
| Krea 2 | Krea AI | Image & Style Transfer | Style‑focused model that competes on aesthetics—illustration, anime, photorealism—rather than just content. |
| Void | Netflix | Video Object Removal | Removes subjects and all associated artifacts (shadows, reflections) in a single pass; Apache 2.0 licensed. |
| Tripo 3.1 | Tripo AI | 3D Generation | Text‑to‑model, image‑to‑model, and multiview‑to‑model in one node; pairs with TripoSplat for image‑to‑3D‑Gaussian pipelines. |
| Luma UNI‑1 | Luma AI | Image Editing | Decoder‑only autoregressive transformer; up to 9 reference images, stronger prompt reasoning, precise editing. |
| Claude | Anthropic | Multimodal Language Model | Adds prompt writing, workflow reasoning, and multimodal understanding directly into the node. |
| OpenRouter | – | Text LLM Hub | Single node providing access to 20+ LLMs; route the best model to the right task without workflow changes. |
| Gemma 4 | Google DeepMind | Multimodal Model | Text, image, audio, video support; runs on a phone or single GPU; 31B version ranked #3 on open Arena leaderboard. |
| HidDream‑O1‑Image | Hidream.ai | Reasoning‑Guided Image Generation | Pixel‑level Unified Transformer (UiT) without external VAEs; excels at complex compositional prompts. |
| Stable Audio 3 | Stability AI | Text‑to‑Audio | Produces music, sound effects, and production‑ready audio ready for inclusion in audio‑centric workflows. |
| BiRefNet | CAAI AIR | Background Removal | High‑resolution background segmentation; complements Void for stills versus video. |
| MoGe | Microsoft | 3D Geometry & Depth | Single‑image full‑3D geometry, depth, normals, and FOV in one forward pass—CVPR 25 Oral. |
Why This Matters for Workflow Creators
-
Expanded Creative Toolkit
- Creators can now mix model strengths (e.g., use Krea 2 for style transfer and Luma UNI‑1 for precise edits) without leaving ComfyUI.
-
Multimodal Synergy
- Claude and Gemma 4 provide language reasoning and multimodal understanding that can drive decisions within a graph—ideal for dynamic, conditional workflows.
-
Audio Integration
- Stable Audio 3 opens the door to complete audio‑visual projects: text‑to‑audio synchronised with video frames from Void or Tripo 3.1.
-
3D & Geometry Workflows
- Tripo 3.1 and MoGe give template authors an entry point into 3D model generation and depth estimation, enabling richer renders or AR applications.
-
One‑Stop LLM Access
- OpenRouter’s single node eliminates the need to manage multiple LLM connectors, simplifying both template development and end‑user maintenance.
How This Affects ComfyUITemplates.com Templates and Apps
-
Template Richness
- Template authors can now bundle multiple models (e.g., Krea 2 + Luma UNI‑1 + Void) into cohesive, reusable graphs, improving the quality and diversity of listings on the directory.
-
Searchability & Tags
- ComfyUITemplates.com can enrich metadata by tagging templates with the new model names, helping users quickly locate workflows that leverage, for example, “Gemma 4” or “MoGe”.
-
Versioning & Compatibility
- The directory should note the required ComfyUI version for each new node, ensuring that users importing templates avoid runtime errors.
-
Community Labs
- With ComfyHub crossing 500 workflows, the directory can curate “May Highlight” collections—templates that showcase novel combinations made possible by the new releases.
Further Reading
- Full announcement on the ComfyUI Blog: ComfyUI May Wrapped
- Explore existing workflows on the community hub: ComfyHub
ComfyUITemplates.com remains committed to helping creators discover, share, and refine workflows that leverage the latest advances in the ComfyUI ecosystem. Keep an eye on the directory for upcoming templates that take full advantage of these new models.