LiblibAI: From Model Community to All-in-One Creation Platform

LiblibAI started as a Stable Diffusion model-sharing community, then upgraded to version 2.0 in October 2025 and became an all-in-one multimodal creation platform — image, video, model training, effects, community, and commercialization, all behind one login.

Core Features

Image models: Seedream 4.0, Qwen-Image, FLUX, MJ V7, 4o-image, and more. Video models: Hailuo 2.0, Kling, Vidu, Tongyi Wanxiang 2.5, Pixverse — swap with one click.

The community is its deepest asset: 100,000+ models and LoRAs covering anime, photography, Chinese traditional aesthetics and more, many unavailable elsewhere. The Inspiration page showcases 20,000+ top-tier AI works — hit "Make Same" to replicate any with one click. Quick Create offers 6,000+ AI application templates, no parameter tweaking needed.

Image generation supports both WebUI and ComfyUI modes. All generated images auto-save to your asset library for direct use in video creation. Model training lets you upload images to train custom LoRAs in the browser — no local GPU required. Post-processing includes resolution adjustment, style transfer, HD upscaling, and inpainting.

Video effects are 2.0's standout feature — particle dissipation, fireworks, on-set one-take, bloom everywhere, dance effects and more. Upload a photo, pick a template, get video with no prompt needed. Effects aggregate capabilities from five video models.

Pricing

Free tier: daily compute for casual exploration, but watermarked and non-commercial. Paid plans split into two tracks: Creation Members (Basic ¥36/month with 15,000 points, Pro ¥63/month with 35,000 points, Master ¥159/month with 90,000 points) and Training Members (starting ¥55/month). Commercial rights require a paid membership.

Who It's For

E-commerce designers for product visuals, social media creators for viral effect videos, AI art enthusiasts exploring 100,000+ models, film teams for creative previs.