Video Gamer is reader-supported. When you buy through links on our site, we may earn an affiliate commission. Prices subject to change. Learn more
Now that Sora has been announced, let’s go over our release date predictions.
Getting access to Sora is extremely limited right now, and though the model has already been revealed, there’s no telling when exactly we’re going to be able to generate realistic and lifelike video to the same effect. As with all of OpenAI’s models, we’re not going to see its public release until testing and development is fully completed.
Sora release date speculation
After being announced on February 15th, the disappointing news is that Sora will likely not release for a while.
As we know, Sora is in its ‘red team’ testing stage. Effectively, the OpenAI team are penetration testing the model and looking for potential exploits and misuses. This is likely one of the reasons there is not a release date set in stone for the model yet.
We’d anticipate that if we’re going to see a release date for Sora, it won’t come for at least three to six months. We’d expect a Sora release date to be with us before August 2024, though there’s nothing to back this up aside from the patterns we’ve seen with previous OpenAI releases.
For example, GPT-4 saw a very soft launch in May 2023, with it’s full launch coming months later and packaged under ChatGPT Plus. Similarly, ChatGPT itself was being worked on and polished for years before its public release in 2022. There’s clearly a precedent for the release of Sora to take a little while.
Why is there no release date for OpenAI’s Sora yet?
We’ve recently seen plenty of controversy surrounding AI with deepfakes being used to impersonate figures in the public eye. This is happening with fairly advanced image generation and basic audio, however Sora’s capabilities seem to far eclipse those. It’s highly unlikely that we will hear of a Sora release date until all the necessary testing has been performed to ensure that no harmful content – whether it’s explicit or misinformative – can be produced.
Guardrails, such as those preventing ChatGPT jailbreaks, are often implemented before release, though more exploits will be found as more users interact with the AI. That’s part of the reason that OpenAI have reached out to artists and creatives to ensure that the Sora model isn’t used to steal, defame, or rip-off their work.
When Sora does release, don’t expect to be able to use it right away
At the moment, it would cost you $0.12 to produce a single HD image in DALL-E 3 at 1792×1024 resolution. We’re going to use this information to predict how expensive it’s going to be generate video with Sora.
Generally, expect to see an FPS of 24 in most natural video. By that logic, it would cost roughly $2.88 to generate a single second of video with Sora, and $172.80 for just a minute of video. That is of course, if the rate of image diffusion to video diffusion matches up, though we expect it to actually be more expensive than this.
The image quality generated by Sora already seems like it far surpasses the realism of DALL-E 3’s potential. Images are sharper, clearer, and lighting is far more lifelike. Generating images to this fidelity is already likely a monumental task, now imagine having to repeatedly do it while also matching up images frame-by-frame. It’s going to be expensive.
Sora doesn’t seem like it’s going to be the type of AI model readily available to the public, unless OpenAI are willing to accept losses on its rendering. Considering the companies recent value rise, that might just be the case.
Frequently Asked Questions
Will Sora cost money when it’s released?
It’s likely that Sora will cost a fair sum when it’s released. It also seems unlikely that it will become integrated into the ChatGPT Plus package, and will likely exist as its own product.
