Definition
Plain language
A Google reasoning system that thinks at length before answering, used for hard problems like olympiad math.
As stated in the literature
A Gemini-based extended-reasoning system from Google DeepMind reported at IMO gold-medal level; cited as one of the frontier-lab math systems that open-recipe efforts like SU-01 position themselves against.
Why it matters: It sets a frontier reference point for deep reasoning that open and reproducible efforts measure themselves against.
For example, faced with a hard olympiad geometry problem, it spends extended time working through the reasoning before committing to an answer.
Heard on the show
“Because up to now the gold-medal-on-IMO story has belonged to a small number of frontier labs — AlphaProof, Gemini Deep Think, OpenAI's reasoning models.”Episode 048 — How a 30B Open Model Reached Olympiad Gold With the Right Recipe