Video & Image Generation


China's video & image generation sector, and every entry we have filed under it.

Chinese labs are among the strongest in video generation, with Kuaishou's Kling, Alibaba's Wan, MiniMax's Hailuo and ByteDance's Seedance competing directly with Western models. Around them a second tier of startups and university labs is releasing open-source models, interactive world models that double as video generators, and production tools aimed at short-video studios, game developers and advertisers, the industries that pay for generated pixels in China.

This page holds our entries on that ecosystem: open-source releases with weights, new versions of AI production workbenches, and research from Westlake, Alaya Lab and others that blurs the line between video generation and world modelling. Consumer app launches are noted when they come with a new model behind them.

Seventeen entries, newest first

Tian Keyu's stealth world-model startup raises $30M from 5Y Capital and IDG at a $200M valuationOct 9 · Funding · 田柯宇Alibaba unveils Qwen4/Qwen5 roadmap and new image, audio, music and world models at Yunqi ConferenceOct 8 · Release · 阿里巴巴HIT Shenzhen and NUS propose QuantWM: training-free 2-bit KV Cache quantization for video world models with up to 6.20x compressionOct 7 · Research result · 哈尔滨工业大学(深圳)iLearn-LabKuaishou's Kling begins preparations for independent Hong Kong IPO targeting at least $1 billionOct 6 · News · 快手科技Yoroll launches two AI-driven interactive romance film-games, turning characters into companion AI agents via Vidu S2 and its in-house H3 SuperfastOct 5 · Release · Video & Image GenerationMiniMax launches public beta of M3.1-Flash, a lightweight model matching Opus 5.5 on frontend generationOct 3 · Release · 稀宇科技(MiniMax)Chinese AI application company Zhiling Xinjing raises tens of millions RMB angel round led by Huace Film & TVSep 28 · Funding · 智灵新境Whale Entertainment launches AI talent program and Jingrui AI Creation Contest with RMB 10M prize poolSep 17 · Strategy · 虎鲸文娱集团RunningHub open-sources MiniMax H3 multi-GPU acceleration stack with ~12x speedupSep 11 · Open source · Video & Image GenerationAnt Lingbo open-sources 1.3B lightweight LingBot-World 2.0 world model that runs in real time on a single GPUSep 10 · Open source · 蚂蚁灵波Wondershare's Wanju Studio launches 'Wanju Going Global Incubation Plan' with over RMB 100 million in resources to support AI film & TV creationSep 8 · Strategy · 万兴科技Westlake University et al. release Code World Model, a code-driven world model paradigmSep 7 · Release · 西湖大学AGI实验室Alaya Lab releases Project Gear to explore next-generation game engineering for AGISep 5 · Release · World ModelsKuaizi.ai launches KP, a unified value unit for video commerce AI, and enterprise KP WalletSep 1 · Release · 筷子科技Xiaoyunque AI announces brand upgrade and creator support with 100 million creditsAug 31 · Strategy · 小云雀AISkyProduction Tiangong Workbench launches a new versionAug 31 · Release · 天工工作台(SkyProduction)Noiz AI Releases HelixWorld 1.0, a Real-Time Interactive Audio-Visual World ModelAug 17 · Release · World Models