OpenAI launches o3-pro as its most powerful and intelligent AI model yet

If you’re a pro user, and ChatGPT feels noticeably sharper or more insightful lately, there’s a reason for that. OpenAI has officially launched o3-pro, a major upgrade to its lineup and the most capable AI model the company has released so far. This update is a definitive shift in how the model approaches problems, using step-by-step reasoning to deliver more accurate and reliable results in complex areas like math, science, and programming.

A step ahead in what AI can do

Think of o3-pro as a next-gen reasoning model. Unlike traditional models that spit out answers in a single pass, o3-pro breaks down problems step by step, especially helpful in domains like physics, coding, or higher-level math. And according to OpenAI, it’s not just better in a one off scenario, it’s consistently better. Internal reviews show o3-pro outperforming the standard o3 model across science, education, business, writing help, and more.

This update comes with practical upgrades too. O3-pro is live in the developer API, priced at $20 per million input tokens and $80 per million output tokens, the latter being the words it generates in response. For context, a million input tokens is around 750,000 words, a bit longer than War and Peace.

Tool access is another big leap. O3-pro can now use Python, browse the web, analyse files, and even interpret visual content, bringing it closer to acting like a true AI assistant. It can remember user preferences across chats too. But all that capability comes with a cost. It’s slower than o1-pro, and some features like temporary chats and image generation are currently off-limits.

ALSO READ: Sarvam Samvaad debuts as India’s first AI assistant for business use across 11 languages

Still, the performance gains are hard to ignore. On the AIME 2024 benchmark (math-heavy), o3-pro edged past Google’s Gemini 2.5 Pro. And in GPQA Diamond, which tests PhD-level science reasoning, it topped Anthropic’s Claude 4 Opus.

This might just mark a new chapter for advanced AI use, especially for researchers, developers, and anyone who wants an assistant that actually reasons. Just don’t expect it to draw you a picture yet.