Sora 2

Headlines · 3

1. Oct 24
   Code Review Challenges a Student AI Startup’s Independence Claims

   虎嗅 AI investigated Ben Zhu, an AI video platform built by a Zhejiang university sophomore who told Hangzhou TV that he earned 600,000 yuan in two months and developed the product almost entirely alone. The author found highly similar code in the GitHub project Infinite Canvas, while Ben Zhu’s main AI requests were routed through a third-party relay and familiar models appeared under product-specific interface names. The student’s revenue and company-formation claims were not independently verified.

   虎嗅 AI · 🔥 0
2. Oct 20
   Changsha's AI comic-drama industry enters a shakeout after rapid expansion

   Economic Observer (author Tian Guobao) reports that Changsha's Malanshan video park once hosted more than 40 AI-generated "comic drama" studios at its peak, and that Hunan government figures disclosed in September 2026 show the Malanshan AIGC-OPC incubator has hosted about 300 entities and released 349 original works. Wei Xu Network's "Du Chongsheng, Shui Zhuang Fu Er Dai" launched on Red Fruit in June 2026 with AI-generated characters, scenes and voice, reaching 12 seasons, six of which entered Red Fruit's AI comic-drama must-watch top ten; the company says it can finish a series in about eight days using the park's "Zhima Manying" platform. Neighboring Guanghe Xingmeng mainly takes production orders from overseas short-drama platforms and is a whitelisted partner of Kunlun Tech's overseas short-drama business; it typically uses five or six people per project, delivering 60-120 minutes in about a month, with around 60 collaborators working as "one-person companies" and splitting revenue by project. One comic-drama producer said only a dozen-plus active studios remain in the park, and Guanghe Xingmeng general manager Wei Lian expects the shakeout to continue into 2027.

   虎嗅 AI · 🔥 0
3. Oct 16
   UBTech says Thinker-WM tops three embodied world-model benchmarks

   UBTech says its in-house unified multimodal world model Thinker-WM ranked first on three embodied world-model benchmarks: WorldOlympiad (overall score 0.740), EWMBench (4.66 total) and WorldArena 1.0 Track-1, claiming it beat models including Qwen-RobotWorld and Sora 2. The model is built around a Unified Dynamics Transformer that encodes RGB video, depth and robot actions into one shared representation space, letting a single set of weights switch between conditioning and generation modes for six capabilities such as joint video-action generation, world prediction and policy reasoning. UBTech says training data comprised roughly 20,000 hours of first-person human video, 10,000 hours of real robot data and 1,000 hours of simulation, and that the model has been validated on its own robot platforms; all figures come from UBTech-supplied material republished with authorization by QbitAI.

   量子位（原生 RSS） · 🔥 0

Experience and discussion from the community

Share my Sora 2 experienceAsk about Sora 2