DeepSeek’s Gold Medal, China’s Answer to OpenAI

DeepSeek doesn’t publish that much research or models these days. But when they do, it’s almost always (or always) worth discussing.

And this time was no exception, as they’ve recently presented the first open-source model to earn the gold medal at the International Maths Olympiad (IMO) 2025, matching OpenAI and Google, who did so a couple of months ago with all the bells and whistles, and placing another AI model’s scores at the level of competitive mathematicians, the best of the best amongst us.

And, luckily for us, unlike the other two, which hide their discoveries from us, DeepSeek has explained to us how they did it, all the while confirming what we already suspected:

There’s no “AI moat”, and the technological difference between US and China models is now measured in weeks.

This is based on reflections I post on TheWhiteBox, where I explain AI in first principles and simple words, for those allergic to hype but hungry for knowledge. For investors, executives, and enthusiasts alike.

Join today for free.