Google DeepMind is a leading AI research lab focused on developing artificial general intelligence (AGI) safely and responsibly, while solving complex scientific problems.
Editorial Standards
These company pages are for readers comparing major AI labs, tracking product strategy, and understanding how research milestones connect to real-world tools and industry influence.
We summarize each company through milestones, products, research areas, and ecosystem impact so the page reflects both what the company builds and why it matters to learners and evaluators.
Last reviewed on March 18, 2026. We refresh timelines and summaries when a company reaches a notable milestone, launches a flagship product, or changes its strategic focus.
Coverage emphasizes notable releases, research significance, product breadth, developer ecosystem relevance, and the company's broader effect on AI adoption and standards.
History
Track the evolution of Google DeepMind's AI technologies and key milestones.
Google DeepMind released Gemini 1.5, featuring a significantly improved context window of 1 million tokens, enabling the model to process and reason over vast amounts of information.
Google DeepMind launched Gemini, a multimodal AI model designed to understand and reason about text, images, audio, video, and code, with versions ranging from Ultra (most capable) to Nano (most efficient).
DeepMind merged with Google Brain to form Google DeepMind, combining two of the world's leading AI research teams under one organization.
DeepMind expanded the AlphaFold Protein Structure Database to include over 200 million protein structure predictions, covering almost all cataloged proteins known to science.
DeepMind introduced Gato, a generalist agent that could perform hundreds of different tasks across a wide range of environments using a single neural network.
DeepMind open-sourced AlphaFold 2, making the groundbreaking protein structure prediction system available to the scientific community.
DeepMind's AlphaFold 2 was recognized as a solution to the 50-year-old protein folding problem, with the results published in Nature.
DeepMind developed MuZero, an algorithm that mastered Go, chess, shogi, and Atari without needing to be told the rules, combining planning with reinforcement learning.
DeepMind's AlphaFold system won the CASP13 protein-folding competition, significantly outperforming other methods and demonstrating a major advance in protein structure prediction.
DeepMind introduced AlphaGo Zero, which learned to play Go without human data, surpassing all previous versions by learning entirely through self-play.
DeepMind's AlphaGo defeated 18-time world champion Lee Sedol in a five-game match of Go, marking a historic milestone in AI development.
DeepMind was acquired by Google for a reported $500 million, while maintaining its research-focused structure.
DeepMind was founded by Demis Hassabis, Shane Legg, and Mustafa Suleyman with the goal of building artificial general intelligence.
