OpenAI removed the waitlist for Sora this morning, fifteen months after the model first leaked in research demos. Anyone with a ChatGPT Plus or Pro subscription can generate video at sora.com. The free tier remains invite-only.
What you get
Plus users ($20/mo) get 50 video generations per month at 720p, capped at 5 seconds. Pro users ($200/mo) get 500 fast generations per month at 1080p, up to 20 seconds, plus unlimited “slow” renders that complete overnight. Both tiers include Sora Turbo, the faster but slightly lower-fidelity variant.
How it compares
Raw output quality remains the best in the category. Physics, lighting, fabric movement — Sora handles them more convincingly than Runway Gen-4 or Kling 1.6. Where it falls behind: editing controls. There’s no motion brush, no green-screen, no character reference workflow. Most professionals will likely use Sora to generate raw footage and bring it into Runway or DaVinci for the actual edit.
The catch
Content filtering is stricter than competitors. We tried generating clips of public figures, branded products, and historical events — Sora refused most of them. Runway and Pika are more permissive. For commercial work involving real-world subjects, that matters.
Sora API access for developers is “later this year”.
