Microsoft-backed OpenAI announced the launch of its ‘Strawberry’ series of AI models, designed to handle more complex reasoning tasks by spending extra time processing queries.
These models represent a significant leap in AI capabilities and aim to solve challenging problems in fields such as science, coding, and mathematics.
The newly introduced models, internally referred to as ‘Strawberry,’ have been released under the names ‘o1’ and ‘o1-mini.’
According to OpenAI, the o1 model is already integrated into ChatGPT and its API. Noam Brown, an OpenAI researcher, confirmed that the models are part of the Strawberry project, which focuses on enhancing reasoning abilities in AI systems.
The o1 model demonstrated significant performance improvements compared to previous versions. In particular, it achieved an 83% score on the qualifying exam for the International Mathematics Olympiad, a substantial jump from the 13% scored by GPT-4o, its predecessor.
Additionally, the model surpassed human PhD-level accuracy on science-related problems and showed superior performance in competitive programming tasks.
