Learn About Google DeepMind's AI Products
Create a free account to access in-depth lessons on each tool and model.
Start Learning Free📋About Google DeepMind
Updated June 15, 2026Google DeepMind is the AI research division of Alphabet (Google's parent company), formed in 2023 by merging Google Brain and DeepMind. Led by Demis Hassabis (Nobel Prize in Chemistry, 2024), the organization is one of the world's premier AI research labs and has responsibility for developing Google's most advanced AI systems.
Google DeepMind develops the Gemini family of multimodal models, currently led by Gemini 3.5 Flash — Google's new default agentic model, posting 76.2% on Terminal-Bench 2.1 with roughly four-times faster output than other frontier models at less than half the cost — and the deeper-reasoning Gemini 3.5 Pro. The lab's research heritage includes AlphaFold (which solved the protein structure prediction problem), AlphaGo (first AI to defeat a world champion Go player), and pioneering work in reinforcement learning, neural architecture search, and multimodal AI. AlphaEvolve, a Gemini-powered coding agent that designs and optimizes algorithms via evolutionary search, has produced documented commercial wins ranging from training-speed improvements at Klarna to materially better DNA sequencing accuracy.
Beyond Gemini and AlphaEvolve, Google DeepMind develops specialized products including Veo (video generation), Imagen (image generation), Gemma (open-weight models), and NotebookLM (AI research assistant), plus agentic products like Gemini Spark (personal always-on assistant), Google Antigravity 2.0 (agent-first developer IDE), Gemini Omni (multimodal image, audio, and text to video), and ERA (Gemini-powered scientific code generator). The lab's distribution footprint is uniquely broad: Gemini is embedded across Google Search, Workspace, Android, and Google Cloud — making the lab uniquely positioned to both advance AI science and deploy it at unprecedented scale.
Google Search has been substantially rebuilt around Gemini-powered agents, with Gemini 3.5 Flash now the default model behind AI Mode worldwide across 98 languages. Gemini's distribution has also expanded beyond Google's own product family through partnerships with General Motors (Gemini Live in roughly 4 million Cadillac, Chevrolet, Buick, and GMC vehicles), the new Googlebooks laptop line (with Acer, Asus, Dell, HP, and Lenovo), and most consequentially Apple (Gemini powers the new standalone Siri app at WWDC June 2026). The Apple partnership makes Google DeepMind a horizontal AI backbone across the Android and iOS device populations, shifting the lab's role from internal-product-only vendor to platform-level AI infrastructure for the consumer electronics industry.
🛠️Products & Tools (18)
Agent-first IDE (VS Code fork). Manager View dispatches 5+ agents simultaneously. Artifacts, multi-model support. Free for individuals.
Google's personal always-on agentic assistant — built on Gemini 3.5 Flash, integrates with Gmail and Workspace, monitors user-set triggers around the clock and dispatches multi-step actions across apps. Announced at Google I/O 2026 and rolling to trusted testers.
Open-source AI coding agent from Google. Routes between Gemini 3.1 Pro and Flash models. 1M token context, plan mode, subagent support, parallel tool scheduling. Free daily limit.
Google DeepMind's Gemini-powered coding agent that autonomously designs and optimizes algorithms via evolutionary search — deployed across quantum, genomics, logistics, and Google's own infrastructure.
Google's cloud storage with 15GB free. AI-powered search, Gemini integration for file Q&A, and native collaboration in Docs, Sheets, and Slides.
Google DeepMind's agentic capability for Gemini 3 Pro and Flash — screenshot-and-click GUI interaction for automating software with no API (preview)
Google's speed-optimized frontier model — 78% SWE-bench, 1M context, ideal for high-volume coding and multimodal tasks.
Google DeepMind flagship model (Feb 2026). 94.3% GPQA Diamond (highest ever), 77.1% ARC-AGI-2, #1 on 12+ benchmarks. 1M token context, native multimodal input (text, image, video, audio), Deep Think reasoning mode. Available via Vertex AI Model Garden and Google AI Studio.
Fast image generation powered by Google's Gemini 3.1 Flash Image model via AI Studio. Speed-optimized for high-quality output with API access for developers.
Google's flagship image gen model — 4K resolution, accurate text rendering, SynthID watermarking.
DeepMind protein-structure prediction AI — has predicted structures for over 200 million proteins, transforming structural biology and drug discovery.
AI features across Gmail, Docs, Sheets, Slides, and Meet. Gemini-powered drafting, summarization, data analysis, and meeting transcription.
Google's AI research assistant. Upload documents, PDFs, or YouTube videos and ask questions grounded in your sources. Generates audio overviews (podcasts) from any content.
Google's multi-step research agent that creates detailed research plans, browses the web autonomously, and produces cited reports in Google Docs format.
Google's multimodal video model unveiled at Google I/O 2026 — turns images, audio, and text into video. The companion piece to Gemini 3.5 Flash, positioned alongside Veo 3 and Nano Banana Pro inside the Google generative-media stack.