DeepMind Blog

Game Arena is a new, open-source platform for rigorous evaluation of AI models. It allows for head-to-head comparison of frontier systems in environments with clear winning conditions.

Deep Think utilizes extended, parallel thinking and novel reinforcement learning techniques for significantly improved problem-solving.

New AI model integrates petabytes of Earth observation data to generate a unified data representation that revolutionizes global mapping and monitoring

We’re publishing a paper in Nature introducing Aeneas, the first AI model for contextualizing ancient inscriptions.

Gemini 2.5 Flash-Lite, previously in preview, is now stable and generally available. This cost-efficient model provides high quality in a small size, and includes 2.5 family features like a 1 million-token context window and multimodality.

Our advanced model officially achieved a gold-medal level performance on problems from the International Mathematical Olympiad (IMO), the world’s most prestigious competition for young mathematicians. It earned a total of 35 points by perfectly solving five out of the six problems.

New experimental AI tool helps people explore the context and origin of images seen online.

Introducing a new, unifying DNA sequence model that advances regulatory variant-effect prediction and promises to shed new light on genome function — now available via API.

We’re introducing an efficient, on-device robotics model with general-purpose dexterity and fast task adaptation.

Explore the latest Gemini 2.5 model updates with enhanced performance and accuracy: Gemini 2.5 Pro now stable, Flash generally available, and the new Flash-Lite in preview.