OpenAI's video generation tool, Sora, is officially here. Now, paying ChatGPT users can generate their own AI video clips using text, photos and even other videos.
CEO Sam Altman announced the public rollout during Monday's 12 Days of OpenAI livestream. The Sora model available today, called Sora Turbo, has been upgraded and is "significantly" faster than the last model the company demoed for researchers in February. Sora's product leaders walked viewers through everything that comes with Sora Turbo -- and there's a lot in the new model.
As with most AI image and video tools, you can enter a text prompt and the program will create short clips based on the ideas you described. With Sora, you can also upload photos and other videos to your prompts as reference material. You can also easily select the video's aspect ratio, resolution and duration, as well as how many variations you want the program to generate. The results are anywhere from 5 to 20 seconds long with resolutions between 480 and 1080p. There are also other post-generation editing features like storyboard, remix and loop for more fine-tuning. If your account is public, all your text-based videos are eligible to be shared on Sora's community Explore page. You can change this in your settings.
Similar to OpenAI's image generator Dall-E 3, you'll need to have a paid plan to access Sora. ChatGPT Plus users (on the $20 per month plan) get 1,000 credits a month, 50 fast/priority generations and are limited to videos of 5 seconds at a max of 720p. ChatGPT Pro, the newest, priciest tier available for power users for $200 per month, eases up on those limits. Pro users get unlimited relaxed generations, longer length and higher resolution options -- and, notably, the ability to download videos without the company's watermark.
Sora is available now in the US, but it's not available in Europe or the UK.
English (US) ·