Adapted by:
Javier Márquez
WriterI've been in media for over a decade, but I've been marveling at the possibilities that technology brings us much longer. I believe we live in a world where the digital revolution is changing everything and that Xataka is the best place to write about it.
Karen Alfaro
WriterCommunications professional with a decade of experience as a copywriter, proofreader, and editor. As a travel and science journalist, I've collaborated with several print and digital outlets around the world. I'm passionate about culture, music, food, history, and innovative technologies.
For months, the AI race had been narrowing. Gemini 2.5 Pro, Claude 4 Opus, and DeepSeek R1 closed the gap with OpenAI’s models by introducing increasingly sophisticated reasoners. The overwhelming dominance of 2022, when ChatGPT left the competition bewildered and several steps behind, no longer defines the current landscape. Today, it’s a head-to-head fight.
However, the company led by Sam Altman hasn’t stood still. It just announced o3-pro, a new reasoning model that aims to reaffirm its leadership position in the industry—at least on paper. It’s available only with specific plans, excluding ChatGPT Plus, which costs $20 per month. Access to the most advanced AI comes at a price.
It offers more tools and context. The o3-pro model builds on o3 but includes an extra stage of reinforcement training and more inference resources. It provides full access to features that make ChatGPT more useful: web search, file analysis, image reasoning, Python usage, and personalization through memory.
These features let users tailor responses to their history, access external sources in real time, and solve complex tasks in multiple steps. It includes:
- Search, code, document analysis, and vision.
- It also offers personalization based on active memory.
- It has a context window of up to 200,000 tokens and a maximum output of 100,000.
More accurate but not faster. O3-pro isn’t designed for speed. Responses take longer than with o1-pro but promise greater accuracy for complex tasks. OpenAI recommends using it when reliability matters more than speed.
It’s a model for those who want things done right, even if it means waiting a few seconds longer. Still, these AI models can produce incorrect information or “hallucinate.”
It theoretically outperforms other models. According to OpenAI’s internal testing, human evaluators preferred o3-pro over o3 in all tested categories: science, education, programming, personal writing, and data analysis.
It also performs better on academic benchmarks. It improves from 86% for o1-pro to 93% in competitive math. It surpasses 2,700 Elo points in programming, compared with 2,517 for o3 and 1,707 for o1-pro.
Available now, but not for everyone. o3-pro is accessible to Pro plan users ($200 per month) and Team plan users, replacing o1-pro in the model selector. OpenAI said Enterprise and educational account users will gain access starting next week.
Access to the most powerful models remains limited to those paying for the most advanced plans. The $20-per-month Plus plan—the most popular among individual users—isn’t mentioned. Everything suggests it’s excluded from this update.
o3-pro is also coming to the API. It will be available not only in ChatGPT but also via OpenAI’s API. That’s good news for those who want to integrate it into products or workflows. Additionally, Altman announced that the price of o3 has dropped by 80%.
The o3-pro API costs $20 per million input tokens and $80 per million output tokens. That’s well above the $2.50 and $10.00 cost of GPT-4o and the new price of o3, now $2.00 per input and $8.00 per output.
Images | Dima Solomin (Unsplash) | OpenAI
View 0 comments