OpenAI Launches Sora 2.0 and Standalone App
OpenAI has released Sora 2.0, a significant update to its text-to-video generation model, alongside a new independent social app. The release integrates video synthesis directly into a conversational interface, a paradigm being called “ChatGPT Time.” Analysts note that the model's ability to simulate physical dynamics and lighting makes it feel more like a "physics engine" for creating plausible video scenarios, rather than just an animation tool.
- The initial public release of Sora was on December 9, 2024, for ChatGPT Plus and Pro users in the U.S. and other select countries, excluding the UK and the European Union. - Sora 2.0, released on September 30, 2025, introduced the capability to generate synchronized dialogue and sound effects along with the video. - The standalone app includes a "cameos" feature, which allows users to insert video of a person into an AI-generated scene, adapting their appearance to the new environment. - The model's architecture is a diffusion transformer, which processes visual data as "patches," enabling it to understand and generate long-range dependencies and maintain object permanence when things are temporarily out of frame. - While Sora can simulate many real-world physics, it has known limitations and struggles with accurately depicting complex interactions like glass shattering, liquids pouring, and correct object state changes, such as a person eating food. - Video generation length is tied to subscription tiers; the free version is capped at 5-10 seconds, the "Plus" tier at 10-15 seconds, and the "Pro" tier at a maximum of 20 seconds. - Access to the full features of Sora 2.0 is primarily through the new iOS app, with an Android version released in select regions in November 2025. - OpenAI offers different pricing tiers, such as the ChatGPT Plus plan ($20/month) for up to 50 videos at 720p resolution and a Pro plan ($200/month) for unlimited videos at 1080p.
