OpenAI has officially released o3 and o4-mini, marking a new milestone in generative AI. These models are described as OpenAI’s smartest and most capable to date. For the first time, users can leverage all tools within ChatGPT—such as web browsing, Python execution, image generation, file analysis, and document interpretation—through agentic reasoning.

This release introduces image reasoning directly into the chain-of-thought mechanism, allowing models to "think" with visual input instead of just referencing it. Users on ChatGPT Plus, Pro, and Team plans will immediately see o3, o4-mini, and o4-mini-high in their selector, replacing the previous o1 and o3-mini variants. Enterprise and Edu plan users will receive access within a week.

The rollout includes full support for developers via the Chat Completions and Responses APIs. These updates are part of OpenAI’s effort to consolidate capability, speed, and flexibility across its AI tools, all while maintaining previous usage rates.