OpenAI has introduced o3-mini, its latest and most cost-efficient model in the reasoning series. Previewed in December last year, o3-mini offers strong performance in science, math, and coding while maintaining low cost and reduced latency, making it an upgrade over OpenAI o1-mini.

This is OpenAI’s first small reasoning model to support highly requested developer features such as function calling, structured outputs, and developer messages, making it production-ready from launch.

Developers can optimize performance by selecting low, medium, or high reasoning effort, enabling o3-mini to “think harder” for complex tasks or prioritize speed when needed. However, the model does not support vision capabilities, with OpenAI o1 remaining the choice for visual reasoning tasks.

Availability and access:

• Developers: o3-mini is rolling out today via Chat Completions API, Assistants API, and Batch API for select users in API tiers 3-5.

• ChatGPT Users: Plus, Team, and Pro subscribers can access o3-mini starting today, with Enterprise access arriving in February.

• Free Users: For the first time, a reasoning model is available to free plan users via the ‘Reason’ option in the message composer or by regenerating responses.

o3-mini replaces o1-mini in the model picker, offering higher rate limits and lower latency. Plus and Team users will now have 150 messages per day (up from 50), while Pro users get unlimited access to both o3-mini and o3-mini-high, the latter providing enhanced intelligence at a slightly slower speed.

Additionally, o3-mini integrates search, allowing users to find real-time answers with linked sources—a prototype feature aimed at future expansion.