How are energy investors positioned?
Friday marked the launch of OpenAI’s latest model, o3-mini, which is now available to developers and users on various platforms. This new model is designed to be the most cost-efficient in the company’s reasoning series, providing exceptional capabilities in science, math, and coding. The o3-mini maintains a low cost and reduced latency, following its preview in December 2024.
The announcement from OpenAI followed upheaval in the AI space this week with China’s DeepSeek launching its open-source AI model, built for a fractin of the cost on older NVIDIA (NASDAQ:NVDA) GPUs.
OpenAI o3-mini stands out as the first small reasoning model from the company that includes developer features such as function calling, Structured Outputs, and developer messages, making it ready for production use immediately. It also supports streaming and offers three reasoning effort options—low, medium, and high—to cater to different use case requirements. These options enable the model to either focus on solving complex challenges or prioritize speed to minimize latency. However, for visual reasoning tasks, developers are advised to continue using OpenAI o1, as o3-mini does not support vision capabilities.
Starting today, subscribers of ChatGPT Plus, Team, and Pro have access to o3-mini. Enterprise access is expected to be available in a week. The new model will take the place of OpenAI o1-mini in the model picker and is set to offer higher rate limits and lower latency. Plus and Team users will see their rate limit tripled, allowing for 150 messages per day with o3-mini, compared to 50 with o1-mini. Furthermore, o3-mini integrates with search functionality to provide up-to-date answers with relevant web source links, currently in early prototype stage.
For the first time, OpenAI is also making a reasoning model available to free plan users through ChatGPT. Users can try out o3-mini by selecting the ’Reason’ option in the message composer or by regenerating a response. This move expands access to OpenAI’s advanced reasoning capabilities to a broader audience.
While OpenAI o1 continues to be the go-to model for general knowledge reasoning, o3-mini offers a specialized option for technical domains that demand precision and quick problem-solving. In ChatGPT, o3-mini uses medium reasoning effort to ensure a balance between response speed and accuracy. All paid users have the choice to select o3-mini-high in the model picker for a version that offers higher intelligence at the cost of slightly longer response times, with Pro users enjoying unlimited access to both o3-mini and o3-mini-high.
Expert testers have evaluated o3-mini and found it to be faster and more accurate than its predecessor, o1-mini, particularly in STEM fields. Testers preferred o3-mini’s responses 56% of the time and noted a 39% reduction in major errors on challenging real-world questions. When set to medium reasoning effort, o3-mini’s performance in difficult reasoning and intelligence evaluations, such as AIME and GPQA, is on par with that of o1.
Yesterday, reports surfaced that OpenAI will be raising up to $40 billion at at a $340 billion valuation. SoftBank (TYO:9984) was said to be in talks to invest up to $25 billion. Microsoft (NASDAQ:MSFT) remains one of the company’s biggest shareholders and partners.
This article was generated with the support of AI and reviewed by an editor. For more information see our T&C.