OpenAI just pulled back the curtain on its most advanced AI system o3-pro, introducing what the company considers its most advanced AI system.
This new model is a leap beyond the usual, designed specifically to handle stepwise reasoning and tackle complex challenges in physics, mathematics, and software development.
Users with a ChatGPT Pro or Team subscription now find o3-pro available, taking the place of its predecessor, o1-pro. Enterprise and academic customers can expect access to roll out within the next week.
OpenAI’s developer API is also welcoming o3-pro, letting businesses tap into its power at twenty dollars for each million input tokens and eighty dollars for each million output tokens.
To put the numbers in perspective, a million input tokens works out to roughly three quarters of a million words.
Expert reviewers within OpenAI’s community have given o3-pro top marks, placing it above the earlier o3 version across a range of critical fields such as teaching, business analysis, scientific research, programming support, and writing assistance.
The upgraded model is praised for improved clarity, accuracy, and its ability to follow directions with greater consistency and depth.
Expanded Abilities and Some Early Hurdles
O3-pro now comes with a toolkit that wasn’t available in older models. It can search the web, analyze uploaded documents, review images, run Python scripts, and even fine tune its responses by remembering context from a conversation.
There is a caveat, though: responses from o3-pro often take longer to arrive, making the experience less snappy compared to o1-pro.
Temporary chat sessions within ChatGPT are not functioning with o3-pro at the moment, due to technical issues under review. Users also won’t find features like image generation or compatibility with OpenAI’s collaborative Canvas workspace just yet.
Despite these early setbacks, performance metrics suggest the model is delivering impressive results.
O3-pro outpaces leading rivals like Google’s Gemini 2.5 Pro on math challenges and surpasses Claude 4 Opus from Anthropic in tests measuring advanced science knowledge.
Internal benchmarks further highlight its ability to respond with detail and thoroughness, adding weight to claims of improved reliability.
Many who have previewed o3-pro are already pointing to its knack for navigating tough domains, signaling that OpenAI’s latest release is not just a minor upgrade but a potential industry standard bearer.
