OpenAI finally releases Sora: 1080p, 20 seconds, and $200 a month

After a much hyped debut in February, OpenAI’s video generator Sora has finally been released with “many limitations.”

There are many odd things about AI business models in these current gold rush days, but few are as eye catching as OpenAI’s final launch of Sora out of research preview mode and into the public arena. Essentially the company is saying: here’s this, it’s not that good yet.

“The version of Sora we are deploying has many limitations,” it writes. “It often generates unrealistic physics and struggles with complex actions over long durations. Although Sora Turbo is much faster than the February preview, we’re still working to make the technology affordable for everyone.”

Affordable is a bit of an issue. Sora is included as part of existing Chat GPT Plus accounts, which currently cost $20 a month. For that you get 1000 credits and the ability to generate up to 50 videos at up to 720p resolution. These are limited to 5 seconds. On the Pro plan, which costs $200 a month, you get 10,000 credits. This provides ‘unlimited relaxed’ videos at up to 1080p resolution, 20s duration, and 5 concurrent generations. They can also be downloaded without a watermark.

Video can be generated in widescreen, vertical or square aspect ratios. Your own assets can be used to extend, remix, and blend, or you generate entirely new content from text.

This all represents quite a jump in capability from the February initial preview. Sora also has a new UI that has been designed to make it easier to prompt it with text, images and videos. A storyboard tool lets users precisely specify inputs for each frame. All Sora-generated videos come with C2PA metadata, which will identify a video as coming from Sora and can be used to verify origin. Uploads of people will also be limited at launch, but OpenAI intends to roll the feature out to more users as we refine our deepfake mitigations

So yes, it’s not that good yet, is limited, OpenAI is still being very evasive about the data it’s been trained on, and it will still take you on a wild ride into the uncanny, the improbable, and the plain unpleasant at the drop of a hat. Furthermore its distribution is limited too. It is also not currently available to people under the age of 18, and while the company says users can access Sora everywhere ChatGPT is available, that also currently excludes the United Kingdom, Switzerland, and the European Economic Area. “We are working to expand access further in the coming months,” it says.

But for all that, and it is a lot, it is still news. Sora is doing something genuinely new and, presumably, will only get better at it. Releases like this serve to remind investors that the gold rush is very much still underway and that the race for the big claims is still wide open. 

tl;dr

• OpenAI has released its video generator, Sora, out of research preview mode, acknowledging that it has "many limitations" including unrealistic physics and challenges with complex actions.
• Sora is integrated into existing ChatGPT Plus and Pro plans, offering limited video generation capabilities at varying subscription costs.
• Users can generate videos in multiple aspect ratios using their own assets or create content from text, and the platform includes a new user interface for easier interaction.
• Access to Sora is restricted for users under 18 and is currently unavailable in the UK, Switzerland, and the European Economic Area, though OpenAI plans to expand access soon.