ChatGPT maker OpenAI developing new breakthrough reasoning technology code-named ‘Strawberry’. Why is it important?

Sam Altman-led OpenAI is working on a new reasoning technology for its large language models (LLMs), codenamed strawberry, Reuters reported on Friday, citing internal company documents and people familiar with the matter. The ChatGPT maker is reportedly hoping that Strawberry will dramatically improve the reasoning capabilities of its AI models.

According to the report, Strawberry is a closely guarded secret even within the organisation. It was previously known as Q* and was seen as a breakthrough within the company.

Also Read | OpenAI unveils' Five-Tier' system to gauge AI progress towards human surpassing abilities: How it works

However, OpenAI has shown Q* demos to some of its staff which show the LLMs capable of answering tricky science and maths questions that are currently beyond the reach of commercially available models.

The document describes a project using Strawberry models to enable the company's AI not only to generate answers to queries, but also to plan ahead enough to navigate the internet autonomously and reliably to perform what OpenAI calls "deep research", according to the source.

Also Read | Apple to get OpenAI board observer seat for integrating ChatGPT in iOS 18. Everything we know so far

What is Strawberry? 

Strawberry will reportedly mark a specialized way of processing an AI models after it has been pre-trained on very large datasets. It includes a specialized way of ‘post-training’ OpenAI's generative AI models or adapting them in order to improve their performance in specific ways even after they have been ‘trained’ on generalized data.

OpenAI reportedly wants to use Strawberry for performing long-horizon tasks (LHT), which require an AI model to plan ahead and perform a series of actions over an extended period of time. 

Specifically, OpenAI wants its models to use these capabilites for conducting research by browsing the web autonomusly with the support for ‘computer using agent’ or CUA which will be able to take action based on its findings.