The 2025 International Olympiad in Informatics (IOI) is one of the world’s most prestigious programming competitions.
What is the 2025 International Olympiad in Informatics?
It gathers the brightest young human coders from around the globe to solve complex algorithm and problem-solving challenges. Participants are given a strict time limit and must complete programming tasks without external help, internet access, or advanced coding libraries.
This year, the competition also included an AI track—a separate category where artificial intelligence systems could compete under the same rules as humans.
According to OpenAI research engineer Sheryl Hsu, the AI system entered the online AI track of the IOI, competing with the same five-hour time limit and 50 submission cap as human participants.
The system ran without internet access or retrieval tools and had only basic programming capabilities—just like the human contestants.
“We scored higher than all but 5 of the 330 human participants and placed first among AI participants,” Hsu said.
The AI used an ensemble of OpenAI’s general-purpose reasoning models. No model was specially trained for the IOI. Instead, the system relied on a strategy to choose which solutions to submit and how to interact with the IOI’s scoring system.
This marks a major improvement from OpenAI’s 2024 attempt, when it finished just below a bronze medal. In one year, it jumped from the 49th percentile to the 98th percentile.
Winning gold at the IOI shows that OpenAI’s models can handle extremely complex, logic-heavy tasks that require reasoning, not just memorisation. It also comes just a week after OpenAI launched GPT-5, its most advanced generative AI model to date.
GPT-5 has been rolled out to all ChatGPT users—both free and paid—and is designed to be smarter, faster, and more reliable than its predecessors. OpenAI says it performs at an “expert” level in fields like mathematics, science, finance, and law, and is far less likely to produce incorrect answers.
While the IOI competition didn’t use GPT-5 directly, the methods and reasoning systems behind it are closely related. OpenAI says this victory, along with recent successes at other programming and maths competitions, reflects how rapidly its technology is improving.
The AI’s gold medal win is an example of how cutting-edge generative AI can move beyond text generation into high-level problem solving—a skill that could be used in software engineering, research, and complex decision-making.