OpenAI Rolls Out New 'Strawberry' AI Models For Advanced Problem Solving
OpenAI says that these models can independently break down complex queries, enabling them to solve harder tasks that previously required user intervention
OpenAI on Thursday launched its latest series of AI models, named “Strawberry,” which are designed to excel in reasoning tasks and solve complex problems in science, math, and coding.
The new models, named “o1” and “o1-mini,” represent a significant leap in performance compared to earlier versions, such as GPT-4o. OpenAI says that these models can independently break down complex queries, enabling them to solve harder tasks that previously required user intervention.
The o1 model’s capabilities are particularly impressive, scoring 83 per cent on the challenging International Mathematics Olympiad (IMO), a dramatic improvement over GPT-4o, which only managed 13 per cent.
These models also outperformed previous iterations in competitive coding environments, reaching the 89th percentile in Codeforces competitions, showing their enhanced capacity to handle advanced computational tasks. OpenAI’s strategy in developing the o1 model involved using a chain-of-thought prompting technique, allowing the model to think through problems in a way that mirrors human reasoning.
The new models have been made available through ChatGPT and its API, with the preview and mini versions accessible to ChatGPT Plus subscribers.
“As an early model, it doesn’t yet have many of the features that make ChatGPT useful, like browsing the web for information and uploading files and images. For many common cases GPT-4o will be more capable in the near term…But for complex reasoning tasks this is a significant advancement and represents a new level of AI capability,” OpenAI mentioned on its blog.
Availability
Starting today, ChatGPT Plus and Team users can access the o1 models in ChatGPT. Users will have the option to manually select either the o1-preview or o1-mini models through the model picker.
Initially, usage will be limited to 30 messages per week for o1-preview and 50 messages per week for o1-mini, though there are plans to raise these limits.
Additionally, OpenAI is working on enabling ChatGPT to automatically select the most appropriate model for each prompt.
The AI company said that ChatGPT Enterprise and Edu users will get access to both models beginning next week. The o1-mini is expected to be extended to free-tier users at a later stage, broadening access to these cutting-edge AI tools.
