OpenAI is reportedly working on more advanced AI models capable of reasoning and ‘deep research’

A new report from claims OpenAI is developing technology to bring advanced reasoning capabilities to its AI models under a secret project code-named “Strawberry.” Among the project’s goals is to enable the company’s AI models to autonomously scour the internet in order to “plan ahead” for more complex tasks, according to an internal document seen by Reuters. The project previously went by the name of Q* (pronounced “Q star”), demos of which showed earlier this year that it could answer “tricky science and math questions,” Reuters reports, citing unnamed sources who witnessed the demonstrations.

At this stage, much remains unknown about Strawberry — including how far along in development it is, and whether it’s the same system with “human-like reasoning” skills that OpenAI reportedly demonstrated at an employee all-hands meeting earlier this week, per . But the ability for the company’s AI to conduct “deep research,” as is said to be the aim of Strawberry, would mark a huge leap forward from what’s available today.