Chinese startup DeepSeek has released the first open-source AI model capable of achieving a gold medal at the International Mathematical Olympiad (IMO). The Math-V2 model reached the gold-level score at this year's competition, a feat accomplished by only 8 per cent of human participants, by demonstrating reasoning abilities rather than simply producing answers.
DeepSeek has made the model available on developer platforms Hugging Face and GitHub, allowing anyone to run or modify it for free. Hugging Face co-founder and chief executive Clement Delangue said: “Imagine owning the brain of one of the best mathematicians in the world for free to explore it for research, fine-tune it, optimise it, run it on your own hardware. No limitations, no nerfing, no company or government to take it back. That’s democratisation of AI and knowledge at its best.”
Models built by Google DeepMind and OpenAI have also achieved gold-level performances at the IMO, but neither company has made them open-source. DeepMind’s model is available only to subscribers of its premium Ultra package, while OpenAI has not publicly released its model.
DeepSeek’s model proved its ability to “self-verify”, according to the firm’s researchers, allowing it to solve mathematical problems without known solutions. This addresses a key bottleneck in current AI systems, which tend to improve only on tasks with easily verifiable solutions.
DeepSeek launched earlier this year, making headlines for claims that its large language model was created at a fraction of the cost of rivals like OpenAI’s ChatGPT. It quickly became the most downloaded free app in the US, leading President Donald Trump to describe it as a “wake-up call” for the tech industry. However, several states have banned the app on government devices amid data security concerns, and Italian regulators have ordered a nationwide ban, citing potential data breaches by DeepSeek’s parent companies.
