Mortal Domains

The full version of OpenAI’s latest AI model, called 01, appears to have leaked on Friday – only for the company to shut it down just two hours later.

As Tom’s Guide reportsa number of users on X-formerly-Twitter discovered that a simple modification of the URL allowed them to access the AI ​​model.

The model was first announced in September and was only available in “preview” form to paying users since then.

But by changing the URL, users claimed to have found a workaround to access the full content.

In a statement to FuturismAn OpenAI spokesperson confirmed that “we were preparing for limited external access to the OpenAI o1 model and encountered an issue.”

“This has now been corrected,” the statement said.

The previews users have gotten so far suggest that the full release could mark a serious improvement over anything we’ve seen previously from the company.

Let it rip

Users were initially impressed by the capabilities of the alleged model, solve a complex math problem to a picture puzzle.

A user found that the AI ​​model could spit out a “complete thought chain 01” after being asked to analyze a photo from a recent SpaceX launch.

Another Reddit user find that it “managed to process a massive JSON dump that was not feasible with o1-preview due to its token limitations”, referring to a common file format that coders used to store human-readable text. ‘man.

We still don’t know when OpenAI will make the full version of its o1 model available to users. As Tom’s guide points out, the company led by Sam Altman may be waiting for the end of the current US presidential election this week.

But even in its “preview form,” the o1 model has already impressed experts with its improved ability to solve standardized tests and new chain of thought reasoning.

Despite some impressive benchmarks, OpenAI recently found that its ability to provide correct – and not “hallucinated” – answers still leaves a lot to be desired.

Whether this will change with the public release of the full version of the o1 model remains to be seen.

Update: The article has been updated with a statement from OpenAI.

Learn more about OpenAI: OpenAI research reveals that even its best models give wrong answers much of the time