Google DeepMind AI releases & news

Every Google DeepMind AI release AI/TLDR has verified from a primary source: 35 so far, mostly 22 models, 3 papers and 3 ecosystem. Latest: Genie 4 — DeepMind's world model adds native audio and live text steering (2026-10-27).

35 releases tracked · deepmind.google

• Genie 4 — DeepMind's world model adds native audio and live text steering2026-10-27 · model · major

  A world model you can steer with words while it runs, now with sound, speaking characters and simulated physics.

• Forest Typology 2020 — DeepMind maps six forest types at 10 m worldwide2026-10-22 · dataset · notable

  An open map that tells primary forest apart from plantations, for every 10 m patch of land on Earth.

• SynthID Bio — DeepMind watermarks AI-designed proteins without breaking them2026-09-30 · paper · major

  A hidden, checkable signature inside AI-designed proteins that does not change what the protein does.

• Gemini 4 Argon — Google's new frontier model gets a 1M-token output limit2026-09-30 · model · seismic

  Google's answer to Opus 5.5 and GPT-6 Astra, built for long coding, legal and finance jobs and for cyber defence.

• Gemini 3.8 Live with Live Avatar — Google's voice agents get a talking face2026-09-24 · model · major

  Gemini 3.8 Live can now appear on screen as a lip-synced video avatar that talks, listens and uses tools in real time.

• Gemini 3.8 Live — Google's voice models talk while they think2026-09-15 · model · major

  Google's new Live API models hold a real-time voice conversation and keep reasoning in the background while they speak.

• AlphaGenome Atlas — a prediction for every possible DNA letter change2026-09-08 · dataset · major

  AlphaGenome's predictions for every single-letter DNA change, precomputed so researchers can look one up instead of running the model.

• WeatherNext 3 — DeepMind's weather model goes hourly at 5km2026-09-03 · model · major

  DeepMind's global forecast model moves from 6-hour, 25km predictions to hourly ones at 5km.

• Gemini 3.8 Flash — Google's best coding model yet, plus a cyber twin2026-09-02 · model · major

  Google's new Flash model keeps the 3.7 price while scoring higher on coding, reasoning and security work.

• Double-blind AI evaluations — DeepMind tests Gemini without seeing the prompts2026-08-27 · benchmark · major

  Google DeepMind ran the first double-blind evaluation of a frontier model: neither side could see the other's secrets.

• Gemini 3.5 Transcribe — Google's speech model cleans up your ums and ahs2026-08-26 · model · major

  Google's new speech-to-text model writes the sentence you meant, not the one you stumbled through.

• Gemini 3.7 Flash — Google's coding and agent workhorse at half the price2026-08-13 · model · major

  Google's new Flash model gains 16 points on long-horizon coding work and launches at half the price of the model it replaces.

• SL2T — Google's sign language model turns signing into text on Pixel2026-08-12 · model · major

  Google DeepMind's SL2T reads a signer's body landmarks and writes the text, and it now runs inside Gboard and Live Transcribe.

• WeatherNext Cyclones — DeepMind's Nature-published model adds a day of warning2026-08-06 · model · major

  A neural cyclone forecaster that matches operational models one day sooner and ships open under Apache-2.0.

• Gemini Robotics ER 2 — the planning brain that watches video and coordinates robots2026-07-30 · model · major

  DeepMind's embodied-reasoning brain plans multi-step robot work, watches the video, and orchestrates multiple robots at once.

• Gemini Robotics 2 — DeepMind's new humanoid model controls the full body2026-07-30 · model · major

  DeepMind's new humanoid brain walks, crouches, and ties knots — and coordinates two robots at once.

• Lyria 3.5 — Google DeepMind's new music model in free Flow Music2026-07-29 · model · major

  Google DeepMind's new music model — richer melodies, clearer vocals, longer songs in the free Flow Music app.

• Gemini 3.6 Flash — Google's workhorse plus Flash-Lite and Flash Cyber2026-07-21 · model · major

  Google ships three Flash-tier Gemini variants tuned for coding, high-throughput agents, and cybersecurity.

• Gemini 3.5 Flash gets Computer Use — native browser, mobile, and desktop agents2026-06-24 · model · major

  Gemini 3.5 Flash can now see a screen and click, type, and scroll on its own through a single built-in tool.

• DeepMind Opens $10M Multi-Agent AI Safety Research Grant2026-06-11 · ecosystem · major

  A $10M research call for tools to keep millions of AI agents safe when they start talking to each other.

• Google Ships DiffusionGemma — Apache-2.0 26B/3.8B-Active Mixture-of-Experts That Denoises 256 Tokens in Parallel via Discrete Block Diffusion, Hits 1,000+ Tokens/Sec on H100 and 700+ on RTX 5090 While Posting 77.6% MMLU Pro, 73.2% GPQA Diamond, and 69.1% LiveCodeBench v62026-06-10 · model · major

  Google opens a 26B-parameter Gemma that denoises 256 tokens at once instead of generating them one by one.

• Google DeepMind Ships Gemini 3.5 Live Translate — Audio Model Streams Near Real-Time Speech-to-Speech in 70+ Languages With Preserved Intonation, SynthID Watermarks, and Public Preview on Gemini Live API, AI Studio, Meet, and Google Translate2026-06-09 · model · major

  First production speech-to-speech model that translates continuously instead of taking turns.

• Google Ships Gemma 4 QAT Checkpoints — New Mobile Quantization Format Brings Gemma 4 E2B Under 1GB of RAM, Q4_0 Weights Land for E2B, E4B, 12B, and the 26B MoE on Hugging Face2026-06-05 · model · major

  Gemma 4 shrinks to under 1GB on phones via a custom 2-bit mobile quantization format trained with QAT.

• Project Genie Plugs Into Street View — Google's World Model Generates Interactive Stylised Scenes Anchored to Real US Locations, Rolling Out to AI Ultra Subscribers Globally2026-05-19 · model · major

  Google's Genie world model now takes a real Street View address as its scene seed — pick a styled lens, walk around a generated version of that exact block.

• Google DeepMind Co-Scientist Lands in Nature — Multi-Agent Gemini System Generates, Debates, and Evolves Scientific Hypotheses With 100+ Research Partners2026-05-19 · paper · major

  Seven Gemini agents argue, rank, and refine scientific hypotheses — now a Nature paper and a Google Labs tool.

• Gemini 3.5 Flash — Google's I/O 2026 Model Rivals Flagship Frontier Models on Coding and Agentic Tasks at 4x the Speed2026-05-19 · model · seismic

  Google's new Flash-tier model targets agentic coding at flagship quality and 4x the throughput.

• Google DeepMind Takes Minority Stake in EVE Online Maker — CCP Spins Out as Fenris Creations in $120M Deal2026-05-06 · ecosystem · major

  DeepMind buys a minority stake in EVE Online's studio and turns the 22-year-old MMO into an offline lab for studying long-horizon AI agents.

• AlphaEvolve Year-in-Review — DeepMind's Gemini Coding Agent Lifts AC Power Flow Feasibility 14% to 88% and Cuts Spanner Write Amp 20%2026-05-07 · tool · major

  DeepMind's AlphaEvolve has graduated from research demo to producing measurable improvements across genomics, grid optimization, quantum hardware, and Google's own infrastructure.

• Google DeepMind UK Workers Vote 98% to Unionize — Cite Pentagon Gemini Deal and Israeli Military Use2026-05-05 · ecosystem · major

  DeepMind's London staff become the first frontier-lab workforce to formally pursue union recognition.

• DeepMind AI Co-Clinician — Talker/Planner Dual-Agent Hits Zero Critical Errors on 97 of 98 Primary Care Queries2026-04-30 · paper · major

  DeepMind's clinical AI uses a Planner agent to keep a Talker agent inside safe medical boundaries during live audio/video patient consultations.

• Gemini Robotics-ER 1.6 — Instrument Reading Jumps from 23% to 93% with Agentic Vision2026-04-14 · model · major

  Gemini's robotics reasoning model now reads physical gauges with 93% accuracy — up from 23% in the previous version.

• Decoupled DiLoCo: 236× Less Bandwidth for Distributed LLM Training2026-04-23 · algorithm · major

  Google DeepMind cuts inter-datacenter training bandwidth 236× while maintaining 88% goodput when chips fail — validated at 12B scale.

• Google Deep Research Max — Autonomous Research Agent Hits 93.3% DeepSearchQA with MCP Support2026-04-21 · tool · major

  Google's Deep Research Max is an async research agent that plans, iterates, and synthesizes — now with MCP access to private data and native chart generation.

• Project Genie — Google DeepMind's Interactive AI World Generator2026-01-29 · showcase · major

  Type what you want to see and walk around inside it — a text-to-explorable-world model running in real time.

• Gemma 4 — open models for reasoning and agents2026-04-02 · model · major

  Four open models from Google under Apache 2.0 that bring frontier-tier reasoning and native tool-use to consumer GPUs and phones.

← All releases · Learn AI