OpenAI is developing an innovative approach to its artificial intelligence models under the project codename “Strawberry,” according to sources familiar with the matter and internal documents reviewed by Reuters. The project aims to demonstrate advanced reasoning capabilities in AI models offered by the Microsoft-backed startup.
Teams within OpenAI are actively working on the Strawberry project, as detailed in recent internal documents seen by Reuters in May. These documents outline plans for utilizing Strawberry to conduct research autonomously, potentially navigating the internet and performing what OpenAI terms as “deep research.”
Details about how Strawberry operates remain closely guarded within OpenAI, according to the sources. The project focuses on enhancing AI’s ability to not only generate responses but also to plan ahead and navigate complex tasks over extended periods, a capability crucial for achieving human-like reasoning.
OpenAI’s efforts with Strawberry also include training models on a specialized “deep-research” dataset, enabling them to autonomously browse the web and potentially assist in software development and engineering tasks. The project underscores OpenAI’s commitment to advancing AI’s reasoning abilities, aiming to surpass current limitations in AI technologies.
The development of Strawberry represents a significant step in OpenAI’s quest to enhance AI capabilities and is part of broader industry efforts to improve AI’s ability to handle complex real-world problems and tasks.
