An advanced version of Gemini with Deep Think achieved a gold medal at this year's IMO! 🥇 So many amazing contributions from the team. What's next? deepmind.google/discover/blog/…

Excited to share that a scaled up version of Gemini DeepThink achieves gold-medal standard at the International Mathematical Olympiad. This result is official, and certified by the IMO organizers. Watch out this space, more to come soon! deepmind.google/discover/blog/…

So happy to see this incredible achievement. Huge congrats to Thang Luong, Quoc Le, Yi Tay and the IMO team on the result. This was a great collaboration across teams to build a general Gemini DeepThink model that can also get gold at IMO.

Proud to announce an official Gold Medal at #IMO2025🥇 The IMO committee has certified the result from our general-purpose Gemini system—a landmark moment for our team and for the future of AI reasoning. deepmind.google/discover/blog/… (1/n) Highlights in thread:

Thang did a lot of heavy lifting of this work, thank you!

Kevin is our top notch RL expert, this wouldn't have been possible without his genius

I really enjoyed our slow yet meticulous scientific approach to the IMO work, while we knew in-house that we are 🥇, we waited for the official grading, adhered to all the requirements, had a well-thought out and inclusive announcement. This is what matters in the long run, imo.

I am thrilled to introduce TYDI QA-WANA, a new benchmark for information-seeking question answering. Our dataset includes 28K question-answer pairs across 10 languages of West Asia and North Africa. Key features of TYDI QA-WANA: 1) Questions are genuinely information-seeking, as

The Gemini Deep Think model that achieved IMO gold medal 🥇 is launched! This is a general purpose model that is not only SOTA at math/proofs but also reasoning, code and many others! 🔥 The exact config that achieved IMO gold with scaled up Deep Think is being made available

Yes, we should :)

Further independent verification of 1) strong performance of Gemini models on math, 2) Gemini Deep Think showing very strong results and also clean and well written proofs, 3) move inference time compute -> more wins. We need harder and hard baselines. At some point, it will

I really like the manager style of my good buddy Mostafa Dehghani. Sounds like a lot of fun to be his report. Someday I will aspire to be like him. He is so funny. I remember once he told me he rejected his reports travel because they did not take business and only economy. 😂🫡