The History of Sora: From Research Preview to Sora 2 — OpenAI's AI Video Evolution
From research demo to cinema-quality AI video: the rise of Sora
OpenAI’s Sora set the bar for AI-generated video when it was first shown in early 2024, then moved from research preview to a public product and later to Sora 2. This article traces Sora’s development—from the first preview clips and the launch of Sora Turbo to Sora 2 and Sora 2 Pro—and how each step expanded what’s possible with text-to-video and image-to-video.
Release Timeline & Major Milestones
| Date | Milestone | Significance |
|---|---|---|
| February 15, 2024 | Sora preview | First public preview; 1-minute HD clips; technical report on training |
| December 9, 2024 | Sora launch | Out of research; Sora Turbo at Sora.com; 1080p, up to 20s; ChatGPT Plus/Pro (US/Canada) |
| February 2025 | ChatGPT integration | Plans to integrate Sora into ChatGPT |
| September 30, 2025 | Sora 2 | Sora 2 and Sora 2 Pro; better physics, dialogue/sound; real people in generated scenes; API |
The February 2024 Preview: Sora Goes Public
Sora’s first previews showed what AI video could do at length and in HD
On February 15, 2024, OpenAI unveiled Sora with a set of high-definition sample videos: an SUV on a mountain road, a short fluffy monster, people walking in Tokyo in the snow, and Gold Rush–style footage. The model could generate videos up to one minute long, and OpenAI published a technical report on how it was trained. The preview immediately made Sora a reference point for quality and length in AI video.
Why the Preview Mattered
- Length and resolution: Up to one minute in HD was ahead of most consumer tools at the time
- Scene variety: Realistic scenes, animation, and period-style content in one model
- Research transparency: Technical report gave the industry a clear benchmark to compare against
December 2024: Sora Launches and Sora Turbo
On December 9, 2024, Sora left research preview. OpenAI launched Sora Turbo—a faster variant—as a standalone product at Sora.com for ChatGPT Plus and Pro users in the U.S. and Canada. Videos could be up to 1080p and 20 seconds, with multiple aspect ratios, and prompts could be text, image, or video. Plus subscribers got a set number of 480p videos per month at no extra cost; Pro offered much higher usage.
From Preview to Product
- Speed: Sora Turbo prioritized fast generation for iteration and social content
- Access: Tied to ChatGPT Plus/Pro and a dedicated Sora experience
- Formats: Text-to-video, image-to-video, and video prompts in one interface
Sora 2 and Sora 2 Pro (September 2025)
| Variant | Focus |
|---|---|
| Sora 2 | Speed and iteration; social content and prototypes; faster, lower cost |
| Sora 2 Pro | Production quality; cinematic and marketing use; higher fidelity, longer render |
Sora 2, released on September 30, 2025 to select users in the U.S. and Canada, improved physics accuracy, added synchronized dialogue and sound effects, and could place real people into generated environments while keeping appearance and voice consistent. The API offers two options: Sora 2 for speed and quick iteration, and Sora 2 Pro for high-quality, production-ready output. Both support creation, extension, and remixing of videos via the Video API and the Sora app.
Summary
Sora’s path from a February 2024 research preview to a public product and then to Sora 2 in 2025 shows how quickly AI video has evolved. Understanding this history helps you choose the right tier (Turbo vs Pro) and use Sora effectively for short-form content, prototypes, or cinematic-quality work.
Key Takeaways
- Sora preview (Feb 2024) established OpenAI’s 1-minute HD text-to-video benchmark
- Sora launch (Dec 2024) brought Sora Turbo to Sora.com and ChatGPT Plus/Pro
- Sora 2 (Sep 2025) added better physics, audio, and real-people insertion; Sora 2 Pro targets production use
- Access via Sora.com, ChatGPT, and the Video API (Sora 2 / Sora 2 Pro)
Try Sora on FuseAITools for text-to-video, image-to-video, and Pro workflows in one place.