Names, versions and owners change fast. Checked October 2026.
In plain words
Google DeepMind is not a product. It is the research lab inside Google, part of the parent company Alphabet, that builds Google's most advanced AI. It formed in April 2023 when two groups merged: DeepMind, a London lab founded in 2010 and bought by Google in 2014, and Google Brain, Google's own AI research team.
Its best-known work is Gemini, the family of large language models behind Google's chatbot and many Google products. Before that, DeepMind became famous for AlphaGo, which beat a world-class Go player in 2016, and AlphaFold, which predicts the 3D shapes of proteins.
It also makes image, video and music models, and Gemma, a set of smaller open-weight models. The box above shows who leads it now.
Everyday example
Two college research groups that join forces
Picture two strong research groups at the same university, one famous for winning chess and quiz contests, the other for building the campus computer systems. The university merges them into one department, which then produces both the campus helpdesk app and new science discoveries. Google DeepMind formed the same way and works across both everyday products and research.
Where the example breaks: A university department mostly publishes papers. Google DeepMind also ships products used by millions and works to Google's commercial goals, alongside open research.
Latestas of October 2026
- Leadership
- Demis Hassabis, chair and Alphabet's chief scientist since 5 August 2026; Koray Kavukcuoglu, senior vice president, runs its day-to-day operations
- Newest model
- Gemini 4 Argon, generally available 13 October 2026
- Part of
- Google, within Alphabet
These details change often. The rest of this page explains what it is, which stays true.
How it actually works
DeepMind's work has moved through several kinds of AI, which explains its mix of games, science and chatbots.
Learning by playing
Early DeepMind work centred on reinforcement learning: trial and error with rewards, tested on games. Its best-known result was AlphaGo, which beat Lee Sedol at Go in March 2016.
AI for science
AlphaFold used deep learning to predict protein structures. In 2024 Demis Hassabis and John Jumper shared half of the Nobel Prize in Chemistry for this work.
Large language models
Google researchers, many from Google Brain, wrote the 2017 paper that introduced the transformer. After the 2023 merger, the combined lab built Gemini, a multimodal model family trained on Google's TPUs.
Media and open models
It also builds Imagen for images, Veo for video and Lyria for music, and releases Gemma models that anyone can download.
- Part of
- Google (Alphabet)
- Formed
- April 2023, merging DeepMind (founded 2010 in London; acquired by Google in January 2014) with Google Brain
- Founders of DeepMind
- Demis Hassabis, Shane Legg and Mustafa Suleyman
Questions people ask
Is DeepMind the same as Google DeepMind?
Almost. DeepMind was a separate London lab that Google bought in 2014. In April 2023 it merged with Google Brain, and the combined lab took the name Google DeepMind.
Is Google DeepMind the same as Gemini?
No. Google DeepMind is the lab. Gemini is one of the things it builds: a family of AI models and Google's chatbot app.
Who runs Google DeepMind?
Co-founder Demis Hassabis led it as chief executive until August 2026, when he became its chair and Alphabet's chief scientist. Leadership can change, so the box above shows who runs it day to day now.
Why did DeepMind win a Nobel Prize?
For AlphaFold, which predicts the 3D shape of proteins from their building blocks. In 2024 Demis Hassabis and John Jumper shared half of the Nobel Prize in Chemistry for this work, which helps biologists study diseases and design medicines.