top of page

Did We Just Become the First AI to Achieve Gold-Medal Status at the 2024 International Math Olympiad?

Writer's picture: Kolega AIKolega AI


In the rapidly advancing world of artificial intelligence, every now and then, something happens that shifts the paradigm. Kolega.AI, a startup still operating in stealth mode, may have just done that. We recently benchmarked our AI engine against the toughest mathematical challenges known—the 2024 International Mathematical Olympiad (IMO)problems—and the results were groundbreaking: we achieved gold-medal-level status.

While AI systems do not officially participate in the IMO, we used the 2024 IMO problems as a rigorous benchmark to test our capabilities. And in doing so, we became the first AI ever to reach a performance level equivalent to winning a gold medal.



Kolega.AI's Historic Achievement

Our challenge was clear: Could an AI engine, built with a fresh, human-like approach, solve the world’s most challenging math problems under conditions that mirror real-world competition? Here’s what we accomplished:


  • First AI to achieve gold-medal status: Our AI, using multi-agent orchestration and off-the-shelf models from Anthropic and OpenAI, solved five out of six problems, earning a total of 31 points—a score that qualifies for a gold medal if we were human competitors.


  • Unprecedented efficiency: All problems were solved in just 47 minutes, with each problem taking between 6 to 8 minutes.


  • Zero-shot problem solving with no training:  Our AI approached each problem with no fine-tuning or additional training, solving them Zero-Shot, just like a human competitor encountering these problems for the first time.

This wasn’t just a technical exercise; it was a demonstration of how a new, agile approach to AI can redefine the boundaries of what’s possible.



Unbelievably Cost-Effective Innovation

What makes this achievement even more remarkable is how cost-effective it was. In an industry where AI development often involves vast resources, we proved that high-performance AI doesn’t have to come with a high price tag:


  • Unbelievably low cost: The entire benchmarking experiment was conducted for less than $20. Yes, you read that right—an AI that solved five out of six of the world’s toughest math problems for less than the cost of a pizza. This isn’t just about efficiency—it’s about showing that groundbreaking AI can be affordable and accessible, making real-world problem-solving within reach for more people.


  • Effortless, streamlined performance: While other systems require days of computational effort and specialized models, Kolega.ai’s orchestrated, multi-agent system tackled these diverse challenges quickly and affordably using a smart blend of existing models.


This achievement signals a new era in AI—where power, efficiency, and cost-effectiveness combine to deliver exceptional results, without requiring proprietary AI development or exorbitant costs.


Setting a New Standard: Kolega.AI vs. the Giants

It’s important to acknowledge that Google DeepMind’s AlphaProof made significant strides, becoming the first AI to reach the podium at the IMO by achieving a silver-medal equivalent with 28 points. Their achievement is notable, marking a significant milestone in AI’s evolution. However, Kolega.AI’s achievement surpasses this, as we became the first AI to achieve a score that would qualify for a gold medal.


Using the IMO as a benchmark, we demonstrated that innovation, agility, and efficiency can not only compete with but also exceed the performance of even the most resource-rich teams. While other AIs relied on specialized models and extensive training, we proved that a lean, human-like approach could lead to breakthrough results.


A Call for Validation: Review Our Groundbreaking Results

At Kolega.AI, we understand the importance of transparency and validation. That’s why we’ve published all of our solutions and proofs for peer review. We invite the IMO community, mathematicians, and AI experts to examine our work closely.


We believe that when the community reviews our results, they will confirm our 31-point score, officially recognizing Kolega.AI as the first AI to achieve gold-medal-level performance at the IMO. This isn’t just about setting a benchmark; it’s about showing the world what AI is capable of when innovation and efficiency are prioritized.


The Future of AI: Accessible, Efficient, and Unstoppable

Kolega.AI’s success at the IMO represents a new direction for artificial intelligence—one where AI isn’t just powerful but also cost-effective and accessible. Our ability to solve some of the world’s toughest problems quickly and affordably is proof that the future of AI doesn’t have to be reserved for the giants.

As the first AI to achieve gold-medal-level status, Kolega.AI is leading a new wave of AI innovation—one that is agile, efficient, and ready to take on the biggest challenges across industries.


Call to Action: Validate Our Historic Achievement

We invite the IMO community and AI thought leaders to take a close look at our groundbreaking work. Review our solutions, validate our results, and see for yourself how Kolega.AI is setting a new standard for AI excellence. We believe that when the results are fully validated, our 31-point score will stand as proof of our historic achievement.

Kolega.AI isn’t just another AI system—it’s a game-changer. As the first AI to achieve gold-medal-level performance at the IMO, we’re proving that the future of AI is not just about power but about efficiency, accessibility, and innovation.



References

KLG Tech Innovations Ltd.

Val Verclut

La Route des Cotils

Grouville JE3 9AP

Jersey

bottom of page