May Wrapped

If you blinked, here’s the catch-up.

In May we integrated 11 new models spanning video, image, 3D, audio, and multimodal.

Here’s what’s new.

Krea 2 — Krea AI · Image & Style Transfer

• Krea’s first foundation model, live as a ComfyUI Partner Node on day one. Most image models compete on what’s in the frame. Krea 2 competes on how it looks — style references, moodboards, and aesthetic range across illustration, anime, photorealism, and beyond.

Void — Netflix · Video Object Removal

• Netflix open-sourced VOID (Video Object and Interaction Deletion). Most inpainting tools erase pixels. VOID removes the subject and everything it caused — shadows, reflections, physical interactions. Apache 2.0, natively supported.

Tripo 3.1 — Tripo AI · 3D Generation

• Text-to-model, image-to-model, and multiview-to-model in one node. Pair it with TripoSplat and you have a full image-to-3D-Gaussian pipeline, end to end.

Luma UNI-1 — Luma AI · Image Editing

• Not a diffusion model. Luma’s Uni-1 is a decoder-only autoregressive transformer — it reasons through your prompt before generating. Stronger reference handling, more precise editing. Create and Modify modes, up to 9 reference images. One of the most architecturally interesting image models of the year

Claude — Anthropic · Multimodal

• Claude is now accessible inside ComfyUI. Prompt writing, workflow reasoning, multimodal understanding — language intelligence anywhere in your pipeline.

OpenRouter · Text

• Access to 20+ LLM models through a single node. Route to the right model for the right task without leaving your workflow.

Gemma 4 — Google DeepMind · Multimodal

• Google’s best open model family yet. Text, image, audio, and video — runs on a phone or a single GPU. The 31B model sits at #3 on the open Arena leaderboard. Apache 2.0, 256K context window.

HidDream-O1-Image — Hidream.ai · Image

• Reasoning-guided image generation. Open source model built on a Pixel-level Unified Transformer (UiT) without external VAEs or disjoint text encoders. Strong on complex compositional prompts where most models fall apart.

Stable Audio 3 — Stability AI · Audio & SFX

• Text-to-audio covering music, sound effects, and production-ready audio. Your sound design toolkit, now in ComfyUI.

BiRefNet — CAAI AIR · Background Removal

High-resolution background segmentation. Pairs naturally with VOID — BiRefNet handles stills, VOID handles motion.

MoGe — Microsoft · 3D Geometry & Depth

• Full 3D geometry from a single image — point maps, depth, normals, and camera FOV in one forward pass. A CVPR ‘25 Oral, now in your toolkit.
  

ComfyHub is gaining momentum

• Crossed 500+ workflows on the Hub

• If you haven’t browsed lately, there’s a good chance someone already built the workflow you were about to make.

Stay tuned for June!