DeepMind, known since 2023 as Google DeepMind, is the artificial intelligence research lab of 📝Google and its parent 📝Alphabet, and the developer of the 📝Gemini family of models. It was founded in London in 2010 by Demis Hassabis, Shane Legg, and 📝Mustafa Suleyman, acquired by Google in 2014, and merged with Google Brain in April 2023. Since August 2026 it has been run day to day by Koray Kavukcuoglu as senior vice president, reporting to Alphabet CEO Sundar Pichai, with Hassabis as chairman.
DeepMind built its reputation on landmark research: 📝AlphaGo defeated Go champion Lee Sedol in 2016, and AlphaFold's protein-structure predictions earned Hassabis and John Jumper a share of the 2024 Nobel Prize in Chemistry. Its focus shifted toward products with the launch of Gemini in December 2023, and by 2026 it was shipping a new Gemini Flash model roughly every three weeks; the Gemini app passed 1 billion monthly users on August 11, 2026. On August 5, 2026, Hassabis stepped aside as CEO to become chairman of Google DeepMind and chief scientist of Alphabet, and longtime Google AI leader Jeff Dean left to co-found Discovery Loop, a 📝Public-benefit corporation.
Models
DeepMind's models span language, open-weight, image, video, and audio families; every release is indexed in the 📝Gemini Model Glossary. Status as of September 28, 2026:
- Gemini — the flagship family, from Gemini 1.0 (December 2023) to Gemini 3.8 Flash (September 2, 2026); the most capable Pro model is Gemini 3.1 Pro, still in preview.
- Gemma — open-weight models built from Gemini research, most recently Gemma 4 (April 2026) under the Apache 2.0 license.
- Image — 📝Nano Banana (August 2025) and its successors Nano Banana Pro and Nano Banana 2, which replaced the Imagen line.
- Video — 📝Google Veo 3, Veo 3.1, and Gemini Omni Flash (May 2026), Google's first any-to-any model.
- Audio and music — Lyria 3.5 for full-length songs (September 2026) and Gemini text-to-speech and transcription models.
- Science and world models — AlphaFold, the Genie 3 interactive world model, and Gemini Robotics.
Products
- Gemini app — Google's AI assistant, with 📝Gemini Live for real-time voice conversations.
- AI Mode in Search — Google Search's conversational mode, running on Gemini 3.5 Flash since May 2026.
- Gemini API and Google AI Studio — developer access to Gemini models, including the Antigravity managed agent and Deep Research agents.
- Gemini Spark — a persistent personal agent announced at Google I/O 2026; see 📝Google Launches Gemini 3.5 Flash and Agentic AI Suite at I/O 2026.
Structure
- Ownership — a unit of Google within Alphabet rather than a separate company; its sister company Isomorphic Labs, spun out in 2021, applies AlphaFold-style research to drug discovery.
- Leadership — Koray Kavukcuoglu, SVP, runs Gemini model development, frontier research, and the Gemini app and developer teams; Demis Hassabis is chairman of Google DeepMind and chief scientist of Alphabet. Sources: Axios, Fortune.
- Reach — the Gemini app passed 1 billion monthly active users on August 11, 2026, Google's fastest product to that milestone. Source: Google.