OpenAI is finalizing its next big release. Strawberry, an AI that “thinks,” is coming in two weeks, according to The Information

The world of artificial intelligence (AI) is quite lively these days. The arrival of ChatGPT in November 2022 It was a real earthquake for the tech industry. Many companies updated their plans and started investing millions of dollars to, at best, outperform OpenAI. This resulted in a variety of new algorithm-based products, such as Google’s Gemini and Anthropic’s Claude.

The company led by Sam Altman has not been standing still. It has improved the language models behind its chatbot, and has even announced a promising voice mode that reminds us of the future proposed by the film ‘Her’. Two years after that important initial move, OpenAI could surprise us again. This time, however, with a model focused on reasoning whose launch seems to be very close.

Strawberry, OpenAI’s “secret” model

The name “Strawberry” appeared on the scene in the middle of this year as a result of a series of internal documents to which Reuters had accessThe existence of the project was also confirmed by anonymous sources related to the AI ​​company. At that time there were not too many details about this technology described by some as “A very well kept secret of OpenAI”, although there was talk of great capacity.

One of Strawberry’s strengths, according to sources, would be its ability to reason and autonomously navigate the Internet to come up with its answers, something that internally they would have called “deep research.” Since language models drive all kinds of applications, including ChatGPT, this invites us to imagine an assistant capable of taking over some tasks that we do ourselves, such as searching through web pages.

Now we know a little more about the project. The Information points out that this language model will arrive sooner than previously thought. The information, not confirmed by OpenAI, suggests that the launch will occur “in the next two weeksWhile timelines can always change, this is an interesting update as Strawberry was believed to be landing sometime in the fall on ChatGPT.

The reputable outlet adds that Strawberry will be offered as a standalone offering. We’re not sure what this is, though it’s presumably a model whose access is tied to paying for an additional subscription. Users would find it within the ChatGPT drop-down menu that currently allows us to choose between GPT-4o mini and GPT-4. It’s meant to be advanced, but not useful for everyone.

Rumor has it that by using ChatGPT with Strawberry, the app will have the ability to “think” before giving a response. The model will take between 10 and 20 seconds to carry out the aforementioned research process by consulting sources on the web. As we can see, this model would not be useful for scenarios where quick answers are necessary, although it could be key for those looking for better answers.

If the rumors turn out to be true, the launch of Strawberry will come at a seemingly delicate time for the company. Despite having received more than $10 billion in investment from Microsoft, OpenAI could need huge amounts of money to keep up with the pace of innovation we have seen so far. Developing AI is an expensive task, and Altman is reportedly preparing a new round of investment.

Images | Xataka with Bing Image Creator | Microsoft

At Xataka | Apple is being different with its AI. For better and for worse