Everything we know about Sora, OpenAI's AI video generator

Key Takeaways

Sora turns texts and images into lifelike videos, capturing detailed scenes with crisp textures and smooth motion.
As of now, Sora is still behind closed doors, being put through its paces by red teamers to iron out any safety concerns and make sure it’s ready for everyone to use.
There’s no word yet on what Sora might cost, leaving us wondering about the price tag and if we’ll be able to use it directly within ChatGPT for easy video creation.

San Francisco start-up OpenAI has unveiled an AI video generator that creates videos that look like they’re out of Hollywood. It has released several demos or short videos to show what Sora can do, including breathtaking clips of realistic-looking woolly mammoths walking through snow and a scene of a Tokyo street captured as if by a camera gliding over the city.

OpenAI, the company behind the ChatGPT chatbot and the AI image generator DALL-E, calls its new system Sora after the Japanese word for sky. It is not yet releasing Sora to the public because it’s working with a small group of academics and researchers to understand all its implications first. It announced the model in February 2024 to give a preview of what is coming — so that people can see the capabilities, and so that OpenAI can collect feedback.

7 AI image generators to try: DALL-E isn’t the only option

With the right AI image generator and a simple prompt, you can create anything.

Sora generates videos from text and images

Sora is an AI video generation model developed by OpenAI that can create realistic video (up to 60 seconds long) from text. It is powered by advanced machine learning techniques, specifically a type of generative AI known as a diffusion model, which “generates a video by starting off with one that looks like static noise and gradually transforms it by removing the noise over many steps,” according to OpenAI. It is capable of generating complex scenes with multiple characters, motion, and detailed backgrounds, all based on your prompts.

Example prompts might include:

“A cartoon Kangaroo disco dances.”
“A litter of golden retreiver puppies playing in the snow. Their heads pop out of the snow.”
“Beautiful, snowy Tokya city is butly. The camera moves through the bustling city street…”

One of Sora’s unique capabilities is that it can also generate videos based on still images, animating the contents of the image with attention to detail and realism. Sora uses both publicly available videos and content licensed from copyright holders. Pocket-lint has not yet tested how well Sora works, but OpenAI said it demonstrated impressive capabilities in generating realistic and detailed videos. Like any AI model, it is likely to occasionally produce videos with inaccuracies or unexpected elements, especially in complex scenes.

OpenAI has not said how many videos it’s used to train the system to date. It also has not yet detailed whether Sora supports multiple languages.

How to master GPT-4 in ChatGPT: Prompts, tips, and tricks

I’ve asked ChatGPT-4 to create everything from poetry to a job application. Here’s what I’ve learned.

Sora currently isn’t available for public use

OpenAI has not specified an exact release date. Sora is currently in the hands of select testers (aka “red teamers”) to identify potential risks and ensure it’s safe before wider availability. OpenAI told The New York Times it’s sharing the technology with a small group of academics and other outside researchers who will look for ways it can be misused.

Sora’s pricing isn’t known yet

OpenAI has not yet announced pricing for Sora. The model could follow a similar approach to other OpenAI products, offering a free usage tier with limited access and a paid subscription for more extensive use. It might even be included with a ChatGPT Plus subscription, which currently costs $20 monthly and unlocks access to GPT-4 and DALL-E 3 image generation.

I tried ChatGPT Plus. Here’s everything it can do

ChatGPT Plus is for those who want a reliable, efficient ChatGPT experience. But, for $20 monthly, what else does it include?

You can’t access Sora in ChatGPT right now

Currently, Sora is not accessible through ChatGPT, OpenAI’s generative-AI tool that allows people to generate text, art, code, and more with simple prompts. It launched in late 2022 and has over 180 million monthly users, as of February 2024. Any future integration with ChatGPT would be announced by OpenAI.

Sora’s AI-generated videos are watermarked

Yes, videos generated by Sora are watermarked to indicate they were created by AI. This is part of OpenAI’s efforts to ensure transparency and help distinguish AI-generated content from real footage. Earlier this month, OpenAI announced it is adding watermarks to its text-to-image tool DALL-E 3, but acknowledged that they can “easily be removed.”