Sora 2
2025-09-30 · Publisher : OpenAI · Domain : Artificial Intelligence
OpenAI launches Sora 2, its flagship model for video and synchronized audio generation, with more credible physics and better control, along with an invitation-only social Sora iOS app.
What changes
- Generates video and sound (dialogue, ambient sound, sound effects) with more faithful physics modeling: a missed basketball shot bounces off the backboard instead of teleporting.
- Follows complex instructions across multiple shots while preserving the state of the world; realistic, cinematic and animated styles.
- "Characters" feature: after a video and audio verification recording, you insert your own likeness and voice into a scene.
- Sora iOS app with a feed customizable through natural language, designed for creation rather than consumption, with stronger limits for teenagers.
- Sora 2 Pro, a higher-quality experimental version, reserved for ChatGPT Pro subscribers on sora.com.
### Specifications - Text and image input, video output with audio. - Launch in the United States and Canada, by invitation, free with compute limits; access to sora.com after invitation. - API: sora-2 and sora-2-pro in the Videos API from October 6, 2025 (snapshot sora-2-2025-10-06). - Maximum clip length, resolution and cutoff date: not specified in the post. ### Pricing App free at launch. API (documentation, per second of video): sora-2 at 0.10 dollars in 720p; sora-2-pro at 0.30 dollars in 720p and 0.50 dollars in 1024x1792, a 1080p resolution at 0.70 dollars appears in the current documentation. ### Benchmarks No numerical score published by the vendor; the post relies on generation examples and a system card. ### Key takeaways Sora 2 marks a leap for generative video, with issues of likeness rights and disinformation handled through consent rules. The product has a short lifespan: the Sora app was shut down on April 26, 2026 and the Videos API on September 24, 2026.
Links
Tags : Vidéo, OpenAI, Sora, Audio, Génération vidéo