In a stunning breakthrough, OpenAI’s latest AI model has achieved gold medal-level performance at the International Mathematical Olympiad (IMO), solving five out of six of the world’s most challenging math problems. This milestone marks a significant leap in general AI reasoning, leaving experts in awe.
Unlike specialized AI systems like Google DeepMind’s AlphaGeometry, designed solely for geometry, OpenAI’s model is a versatile large language model that excels in math while maintaining broad reasoning capabilities. The model competed under human exam conditions, mirroring the high-pressure environment of the IMO, held this year on Australia’s Sunshine Coast, where India secured six medals and ranked 7th among 110 countries.
“This isn’t a model trained just for the IMO,” said OpenAI’s Noam Brown on LinkedIn. “It’s a general-purpose reasoning LLM with new experimental techniques, showcasing a major step forward in AI intelligence.” OpenAI CEO Sam Altman called the achievement “a dream come true,” reflecting on AI’s decade-long journey to this point.
The accomplishment stunned even the sharpest minds in mathematics. Terence Tao, an IMO gold medalist and renowned mathematician, had expressed skepticism in a June podcast, suggesting AI wasn’t ready for the IMO’s complexity and should focus on simpler contests. OpenAI’s model has now silenced those doubts, proving its ability to tackle problems that demand deep creativity and logical rigor.
What sets this model apart is its ability to “think” for hours, a significant evolution from earlier models like o1, which processed for seconds, or Deep Research, which took minutes. “This model is not only more thorough but also more efficient in its reasoning,” Brown noted. This extended deliberation mirrors human-like problem-solving, enabling the AI to navigate the IMO’s notoriously intricate challenges.
The achievement underscores the breakneck pace of AI progress. “Just last year, AI labs were benchmarking models on grade school math (GSM8K),” Brown explained. “Since then, we’ve conquered high school-level MATH benchmarks, the AIME, and now the IMO gold.” This rapid trajectory highlights AI’s growing ability to handle abstract, high-stakes intellectual tasks.
The IMO gold isn’t just a trophy for OpenAI—it’s a signal that AI is closing the gap on human-level reasoning across diverse domains. By mastering one of the toughest intellectual challenges without specialized training, OpenAI’s model demonstrates the potential to revolutionize fields requiring complex problem-solving, from scientific research to engineering. As AI continues to evolve, this milestone will be remembered as a turning point in the quest for general intelligence.
Naorem Mohen is the Editor of Signpost News. Explore his views and opinion on X: @laimacha.