AI Shipmas Coming! What Will OpenAI's 12 Days of 12 Announcements Reveal?

After a period of silence, OpenAI is finally going to make a big move. This generative AI giant announced that it will start a 12-day new product blitz starting from Thursday.

12 days, 12 livestreams, a bunch of new things, big and small. 12 Days of OpenAI starts tomorrow, the company said on social media platform X.

And Altman further explained, Each weekday, we will have a livestream with a launch or demo, some big ones and some stocking stuffers. We've got some great stuff to share, hope you enjoy! Merry Christmas.

Text-to-Video Sora Makes a Comeback

Remember OpenAI's text-to-video model Sora released in February this year? Just with a few words of description, it can generate lifelike dynamic videos. However, after making a big splash, this product has been silent, leading many to believe that Sora had been abandoned. But this time, it's coming back.

Several OpenAI employees have been warming up on social media. What's on your Christmas list? a technical staff posted. Got back just in time to put up the Shipmas tree, another staffer wrote. Sora lead Bill Peebles responded to a staffer who posted that OpenAI is unbelievably back with one word: Correct. with IYKYK (if you know, you know).

At the end of November, there was news of the latest version of Sora leaking. What was leaked was not the model weights, but the test interface made public on Hugging Face by relevant personnel. Reportedly, this leak was through a front-end on Hugging Face, where users can generate 1080p 10-second videos by inputting a short text description. But just 3 hours after the project was created, OpenAI shut down all early artist access to Sora.

And OpenAI's former CTO Mira Murati said back in March that Sora is expected to launch by the end of the year, further cementing the blockbuster return of text-to-video. What kind of surprise will it bring this time? Let's look forward to it.

Much-Improved Reasoning Model o1?

OpenAI is also expected to release the complete reasoning model this time. Back in September, the AI start-up released a preview of the latest version of the o1 model, with the ability to handle more complex problems and logical reasoning. However, this model's influence is not as widespread as GPT and Sora, possibly due to the fact that it is still quite early in its use, and the cost is too high (3-4 times the price of GPT-4). Therefore, for this reasoning mega-model to gain more attention, it either needs to achieve a qualitative leap, or significantly reduce the cost.

Little Easter Eggs? Christmas-Exclusive Voice Coming?

OpenAI's so-called little surprises may include integrating Santa Claus's voice into ChatGPT, allowing users to experience the full Christmas surprise interaction. Some GPT users have discovered that the voice mode button in the underlying code has changed to a snowflake shape.

Some netizens are also speculating about new product leaks, such as the new text-to-image model Dell-E4, AI hardware co-developed with Apple's former design chief Jony Ive, self-developed chips, nuclear-related layouts, new features for ChatGPT, and so on. These are all worth looking forward to.

Of course, making predictions now is meaningless, as these new products will be unveiled soon. Let's prepare to welcome the AI Christmas carnival at the end of the year!