Algorithmic trading firm XTX Markets announces the $10M AI-MO Prize for a public AI model than can win a gold medal in the International Mathematical Olympiad

AI isn't smart enough to win a gold medal at the math Olympics—yet. Can a billionaire's money change that? — Alex Gerko cannot wait to lose $10 million.

Wall Street Journal 2023-12-02 Ben Cohen

Context & Ripple Effects

XTX Markets is extending its machine-learning focus beyond trading infrastructure into an open research incentive. Its founder’s later profile described the compute-intensive base behind XTX, including an Icelandic supercomputer and roughly 25,000 AI chips.

The prize fixes a public, difficult benchmark for mathematical reasoning. Later coverage shows why that distinction matters: experimental models were reported at gold-medal level, yet human contestants still outscored leading models at the 2025 IMO.

First-order effects

Researchers and model developers gain a $10 million incentive to build and publicly release systems that can meet a clearly specified Olympiad-level standard.
XTX ties its brand to a technically demanding AI benchmark while directing attention toward public models rather than proprietary-only demonstrations.

Second-order effects

Labs pursuing mathematical reasoning have a new external benchmark against which to compare claimed progress; public release requirements could make qualifying work easier for other researchers to inspect and build on.
The award raises the value of compute, talent, and evaluation methods able to produce verifiable reasoning performance—capabilities already central to XTX’s AI-heavy trading infrastructure.

Third-order effects

If prizes and benchmarks increasingly reward reproducible public performance, AI competition may shift from broad capability claims toward auditable task-level results.
Mathematical competitions can become a recurring proving ground for reasoning models, but later gaps between model and human IMO scores show that benchmark success should not be treated as a complete measure of general intelligence.

The trend: Private AI capital is increasingly using high-visibility, rigorously scored challenges to accelerate and validate progress in reasoning models.

Discussion

@mickeyshaughnes Mickey Shaughnessy on x
Great thread on math AI. It's got a long ways to go, I think. In our Mathcounts team meeting we've been playing with having the LLM solve problems, it's not very good even at the simple ones yet.
@shafayat_sheikh Sheikh Shafayat on x
I saw quite a few people talking about IMO gold medalist model. This is a nice breakdown on the so many things that is still quite far away, specially the spatial understanding is a quite hard one.
@iamtimnguyen Timothy Nguyen on x
As an alternative to solving 5 Clay Math Millennium problems, you can now build an AI that wins a gold medal at the IMO to win $5 million dollars. The AI Mathematical Olympiad Prize: https://aimoprize.com/
@ian_dot_so Ian Tracey on x
This is an awesome contest. State of the art models today are bad at mathematical reasoning. A model that could solve Olympiad math problems would be a step closer to a future of AI-driven scientific discovery https://aimoprize.com/
@alexandergerko Alex Gerko on x
First order of business with AI-MO prize, we are looking for somebody to run this exciting project: https://www.xtxmarkets.com/... If you know someone qualified, please share!
@atabarrok Alex Tabarrok on x
AI Mathematical Olympiad, $5 million prize for an AI that enters math Olympiad and wins gold. https://aimoprize.com/
@wsj @wsj on x
It's the world's most famous high-school math test. Now anyone can win millions of dollars if they ace it—with one catch. https://www.wsj.com/... https://www.wsj.com/...
@_lewtun Lewis Tunstall on x
A wonderful thread on what it takes to do well on the math olympiad and why LLMs will have an uphill battle “solving it”
@rm_rafailov Rafael Rafailov on x
I saw this challenge https://aimoprize.com/ to develop an AI that can win a gold medal at the IMO. I competed at that level a couple of times (only silver medals though) and have been working on RL and LLMs for a bit. Here is my thoughts on what the challenges are: 1/N

Chronicles