AI Model Hub

ACE-Step is an open-source music generation foundation model jointly developed by StepFun and ACE Studio, supporting music generation, editing, and audio synthesis.

2 version pages

Anima is a 2 billion parameter text-to-image model by Circlestone Labs and Comfy Org, focused on anime concepts and artistic styles. Available in Base, Aesthetic, and Turbo variants for ComfyUI.

0 version pages

Boogu-Image is a versatile image generation model series with base, turbo, and edit variants for ComfyUI.

2 version pages

Chroma is a model series by lodestones including Chroma1-HD (8.9B T2I) and Chroma1-Radiance (WIP upscaling), both Apache 2.0 licensed.

2 version pages

ERNIE-Image is Baidu's 8B parameter open text-to-image model series with a light Prompt Enhancer, excelling at text rendering and structured generation.

1 version pages

Flux is an open-source image generation model series by Black Forest Labs. Two generations: FLUX.1 (3.8B DiT) and FLUX.2 (32B dense transformer).

5 version pages

Gemma is Google DeepMind's family of open models for use as ComfyUI text encoders, including Gemma 4 with E2B and E4B instruction-tuned variants.

1 version pages

HiDream-I1 is a 17B parameter text-to-image model by HiDream-ai, combining DiT with MoE architecture. MIT licensed, available in Full/Dev/Fast variants.

2 version pages

Tencent's open-source video generation model series. From v1.0 to v1.5, delivering high-quality text-to-video and image-to-video generation.

2 version pages

Hunyuan3D is Tencent's open-source 3D generation model series. Version 2.0 supports high-fidelity 3D model generation with high-resolution texture maps from text, image, or sketch inputs.

2 version pages

IC-Light is an image relighting model series by lllyasviel. introduces foreground conditioning for professional-quality image relighting.

0 version pages

Ideogram 4 is a state-of-the-art text-to-image AI model by Ideogram AI. Features advanced typography and photorealistic generation capabilities.

1 version pages

DeepSeek Janus Pro is a multimodal AI model capable of both image understanding and generation, open-sourced by DeepSeek in January 2025.

1 version pages

Krea 2 is an open-source AI image generation model by Krea. Features both Raw (full-quality) and Turbo (distilled) variants with style LoRAs for diverse artistic effects.

1 version pages

LongCat by Meituan is a pioneering open-source bilingual (Chinese-English) foundation model for image generation with superior Chinese text rendering and photorealism.

2 version pages

LTX-Video is the first DiT-based video generation model by Lightricks with 2B parameters, supporting text-to-video and image-to-video generation. Open source under Apache-2.0.

3 version pages

Lumina-Image 2.0 by Alpha-VLLM is a 2B parameter flow-based diffusion transformer for text-to-image generation.

1 version pages

NewBie-image Exp0.1 is a 3.5B Next-DiT model by NewBie AI Lab specialized for anime-style image generation.

1 version pages

OmniGen is a unified multimodal generation model series by VectorSpace Lab supporting T2I, image editing, and in-context generation.

1 version pages

Ovis is AIDC-AI's vision-language image generation model series for use with ComfyUI, including diffusion models and text encoders.

1 version pages

PixelDiT by NVIDIA is a VAE-free pixel-space diffusion transformer for text-to-image generation. Includes PiD (Pixel Diffusion Decoder) models for super-resolution up to 4K.

1 version pages

Qwen text encoder models by Alibaba for use with ComfyUI: Qwen3-VL (vision-language) and Qwen3.5 (text only).

2 version pages

Qwen-Image is Alibaba's open-source image generation foundation model in the Qwen series, excelling at complex text rendering and precise image editing.

3 version pages

Stable Diffusion lineage from Stability AI: SD 1.5 (Runway), SD 2.x, SDXL, and SD 3.5. Open text-to-image checkpoints widely used in ComfyUI.

7 version pages

Stable Audio is Stability AI's family of text-to-audio models for music generation, sound effects, and foley, including Stable Audio Open 1.0 and Stable Audio 3.

2 version pages

VOID by Netflix is a video object and interaction deletion model that removes objects from videos along with all induced physical interactions.

1 version pages

Alibaba Cloud's open-source video generation model series. From Wan2.1 to Wan2.2, delivering high-quality text-to-video and image-to-video generation with MoE architecture.

3 version pages

Z-Image is an efficient image generation model family by Tongyi-MAI (Alibaba). Features a 6B parameter single-stream DiT architecture with variants for quality (Z-Image) and speed (Z-Image-Turbo).

2 version pages