Google DeepMind announces that they’ve also achieved a gold medal in the IMO.
They’ve exactly matched OpenAI, getting perfect scores for the first 5 questions and flunking the 6th.
They’re using what sounds like an experimental general version of Gemini which they’re then fine tuning for IMO rather than a maths specific model.
Their solutions were checked by the IMO (unlike OpenAI) and look much more like a neat little mathematics proof instead of the raw scratchpad that OpenAI turned in.
This one is even more surprising than OpenAI’s entry (in its details). Since it can now write proofs well automatically (even if it costs a lot and takes a lot of time), in a few months regular reasoning models might get enough training data to reliably understand what proofs are directly, and that’s an important basic ingredient for STEM capabilities.
I think it is important to note that Gemini 2.5 Pro Capable of Winning Gold at IMO 2025, with good enough scaffolding and prompt engineering.
I wonder if a single model wrote the it’s solutions itself, or if it had a more messy CoT like o3 which humans (or another instance of gemini) translated into latex.