LiblibAI: China’s Largest All-in-One AI Creation Ecosystem Rooted in the Stable Diffusion Community
Within China’s AI art community, LiblibAI (Libu Libu AI) is an indispensable staple for every Stable Diffusion creator. Founded in March 2023, the platform was originally positioned as the GitHub of AI painting — a dedicated sharing hub for original AI generative art models. In just three years, it has grown into China’s largest AI image generation platform, with over 25 million total users, 4 million monthly active users, more than 500 million total generated images, and a daily rendering volume of 5 million graphics. What began merely as a model repository has evolved into a complete all-in-one creator toolkit covering models, image generation, video production and custom production workflows.
Core Moat: Massive Model Repository & Mature Creator Community Ecosystem
LiblibAI seized the opportunity to localize the open-source Stable Diffusion ecosystem, delivering Chinese-language interfaces, fast download speeds, and unrestricted access to AI art models without overseas network tools for domestic creators. Starting with an initial library of over 100,000 models, it now hosts the world’s largest collection of open-source stylistic models, covering illustration, commercial photography, e-commerce visuals, poster design, IP character creation and countless other visual categories.
More crucially, the platform has nurtured and connected over 20 million AI creators, forming a self-sustaining closed-loop ecosystem linking model developers, content creators and enterprise clients. Its B-end partners include well-known organizations such as Kingsoft Office, Wondershare Technology, Gigabit Games, Tmall Campus and Tsinghua University.
October 2025: LiblibAI 2.0 Marks a Critical Upgrade From Separate Tools to a Professional Integrated AI Studio
The milestone 2.0 release unified scattered features — model management, image generation, video creation and custom workflows — within a single dashboard to deliver one-stop creative production. Its unmatched model aggregation capability integrates top mainstream generative models across two verticals:
- Image generation: Qwen-Image, Seedream 4.0, Nano-Banana, Midjourney V7, F.1 and more
- Video generation: Kling 3.0, Hailuo 2.0, Tongyi Wanxiang 2.5, Vidu Q1 and other leading video models
Users gain access to nearly all mainstream AI generative models on a single site, eliminating the hassle of switching between multiple platforms. The updated version also added pre-built video effect templates; uploading a single image instantly generates dynamic motion effects such as takeoff and flight animations, drastically lowering operational barriers for casual creators.
July 2025: StarFlow Agent — China’s Native Design Agent, Paired With Overseas Lovart Agent as a Dual Flagship Product Line
LiblibAI launched StarFlow Agent, its domestic dedicated design intelligent agent, forming a dual-brand layout alongside Lovart, the design agent developed by its overseas subsidiary. StarFlow Agent’s core capability is natural-language-driven end-to-end design: users submit plain-text creative briefs, and the system autonomously completes demand analysis, scheme iteration and final asset output.
In real testing, StarFlow Agent generates a complete preliminary brand VI package within 10 minutes, including logos, color systems, packaging drafts and promotional posters. This capability delivers exponential efficiency gains for e-commerce operators, independent designers and small business owners. The overseas Lovart product performed exceptionally well during beta testing, drawing a waitlist of over 100,000 users in only 5 days, with roughly 200,000 daily active users and an estimated annual revenue run rate of 30 million US dollars.
March 2026: LibTV AI Video Platform Expands the Ecosystem Into Full Video Production
LiblibAI launched its AI video creation platform LibTV in March 2026, integrating mainstream video models including Kling 3.0. It adopts a dual-access architecture supporting both human creators and AI Agents, enabling full end-to-end production from script writing and storyboarding to final rendered footage. The platform surpassed 100,000 visits on launch day, with a team collaboration edition rolled out in May the same year.
With the release of LibTV, LiblibAI’s business scope expanded beyond standalone image generation into a comprehensive AI creation ecosystem encompassing image generation, video production, design Agents and enterprise API services.
Capital Market Recognition: Record-Setting Financing for Domestic AI Applications
In October 2025, LiblibAI closed a $130 million Series B financing round jointly led by Sequoia China and CMC Capital, with participation from Ant Group, Lenovo Ventures and other investors. This represented the largest single financing deal for domestic AI application products in 2025. Prior to this, the company completed four financing rounds within a single year, setting a speed record for China’s AI application track at that time.
LiblibAI follows a differentiated development path compared to ByteDance’s Mengi AI:
- LiblibAI: Focuses on a massive model marketplace + professional customizable workflows, targeting professional creators and enterprise clients
- Mengi AI: Prioritizes self-developed proprietary models + simplified one-click consumer-grade experience, catering to mass casual users
Hybrid Monetization: Freemium Individual Subscriptions + Enterprise Batch API Access
- Free tier: Core basic features with a daily allocation of free compute credits for all regular users
- Basic Membership (¥39/month): 3,000 daily compute credits + dedicated cloud asset storage
- Professional Membership (¥70/month): Unlimited image generation, priority rendering queues and full LoRA model training access
- Enterprise API: Bulk generation capacity supporting up to 200,000 asset renders per month for corporate clients
Exclusive member-only models are also available; visuals generated from these licensed models are cleared for full commercial use. While LiblibAI remains in an investment stage with net losses, its clear profitability roadmap is emerging amid steady growth in paying subscribers and ongoing global market expansion.
Industry Setback & Rectification: Content Safety as an Invariable Bottom Line
In April 2026, CCTV reported vulnerabilities involving pornographic content generation across multiple AI platforms, naming LiblibAI in the coverage. The platform responded immediately by launching full internal audits and comprehensive technical overhauls, optimizing content review logic and model safety guardrails. This incident underscores a universal lesson for all AI platforms: technological advancement must never cross the red line of content security compliance.
Founder Background: Chen Mian, Ex-Global Commercial Lead of Jianying / CapCut
Founder Chen Mian, born in 1992, previously served as the global commercialization director of ByteDance’s Jianying (CapCut). He was one of ByteDance’s youngest Level 4-1 product leaders, an equivalent rank to Alibaba P9. He excels at building closed-loop business systems linking users, content and traffic — a strength clearly reflected in LiblibAI’s growth strategy: enter the market via a model community to accumulate core creator users, extend into integrated production workflows to boost user stickiness, and scale growth via intelligent Agents and global overseas expansion.
In interviews, Chen Mian stated: “The core of application entrepreneurship is to fully anticipate upcoming fundamental shifts in artificial intelligence, and lay groundwork before those transformations arrive.”
Whether you are an AI artist searching for high-quality generative models, an e-commerce operator aiming to streamline design workflows, or a student systematically learning professional AI creative tools, this largest and most complete domestic AI creation platform is well worth visiting on its official website. Join its community of 25 million creators and experience the complete end-to-end creative pipeline from model selection to finished visual assets.