Google DeepMind

• Lyria 3.5 Gen 4 Audio

  Released 8d ago

  Google DeepMind's newest music generation model, launched in Google Flow Music with advances in musicality, lyrics, vocals, and creative control. It creates richer, more natural melodic structures, generates higher qual…

  • 🎵 Music
  • 🎵 Music production
  • 🎵 Lyrics to music New
• Gemini 3.8 Flash Cyber Waitlist Gen 2 New Coding

  Released 10d ago

  Gemini Gemini 3.8 Flash Cyber is a cybersecurity focused variant of Gemini 3.8 Flash, built for vulnerability detection and automated patching. It exceeds 70 percent success on internal vulnerability discovery evaluatio…

  • 🔒 Cybersecurity
  • 💻 Coding
  • 🤖 Agents
  • 🔒 Code security
• Gemini 3.8 Flash Gen 3 Multimodal

  Released 10d ago

  A multimodal model in the Gemini 3 family, built for long horizon software engineering, autonomous agents and complex enterprise workflows. It accepts text, image, audio and video input and returns text, with a context…

  • 💬 Chatting
  • 📊 Data analysis
  • 💻 Coding
  • 🤖 Agents
  • 🔍 Image analysis New
• Gemini Omni 1.1 Flash Gen 4 Video

  Released 16d ago

  A generative video model offering scene extension up to 40 seconds, first and last frame interpolation, 4K upscaling, and fast 360p draft previews for iterative prototyping.

  • 🎥 Videos
  • 🎬 Video editing
  • 📽️ Image to video
  • 🎞️ Video upscaling
  • 🎥 Video extension New
• Gemini 3.5 Transcribe Gen 4 Audio

  Released 17d ago

  A speech to text model offering low latency, high accuracy transcription with speaker identification, word level timestamps, automatic filler word removal, smart formatting, and custom vocabulary support across 85+ lang…

  • 🗒 Transcription
  • 📷 Live captions
  • 🎤 Dictation
  • 📝 Meeting transcription
  • 🎤 Voice transcription New
• Gemini Robotics On Device 2 Waitlist Gen 3 Multimodal

  Released 1mo ago

  Gemini Efficient vision language action model optimized to run locally on robotic hardware, built on Gemini Robotics technology and on device Gemma models, designed for fast adaptation to new robot bodies with minimal t…

  • 🤖 Robotics
  • 🤖 Task automation New
• Gemini Robotics ER 2 Gen 3 Research preview Multimodal

  Released 1mo ago

  Gemini Vision language embodied reasoning model that acts as a robot's high level planner, understanding physical scenes, communicating with humans in natural language, and coordinating multi step tasks and multi robot…

  • 🤖 Robotics
  • 🤖 Task automation
  • 🧩 Spatial intelligence
  • 👁️ Computer vision assistance New
• Gemini Robotics 2 Waitlist Gen 3 Multimodal

  Released 1mo ago

  Gemini Vision language action model that converts visual and language input into motor control, letting robots take physical action. Controls full humanoid robots from feet to fingertips as well as bi arm robots, and ad…

  • 🤖 Robotics
  • 🤖 Autonomous humanoids
  • 🤖 Autonomous cleaning
  • 🤖 Autonomous mobile robots New
• Gemini 3.1 Flash Lite Image Gen 4 Image

  Released 2mo ago

  Gemini Gemini 3.1 Flash-Lite Image, also known as Nano Banana 2 Lite, is Google DeepMinds fastest and most cost-efficient Gemini image model. It generates and edits images at high speed and low cost while preserving the…

  • 📷 Images
  • 🖌️ Image editing
  • ⚖️ Character consistency New
• Gemini 3.5 Live Translate Gen 4 Audio

  Released 3mo ago

  Gemini Gemini Audio is Google DeepMind’s closed-source native audio model family for low-latency live dialogue, controllable speech generation, audio understanding, and voice-first applications.

  • 🎧 Audio translation
  • 🔊 Text to speech
  • 🗣️ Voice cloning
  • 🗒 Transcription
• Magenta RealTime 2 Gen 4 Audio

  Released 3mo ago

  Google DeepMind’s open on-device live music-generation model for low-latency, continuously controllable music from text, audio examples, and MIDI.

  • 🎵 Music
  • 🎵 Music production
• Gemini Omni Flash Gen 3 Multimodal

  Released 3mo ago

  Google’s multimodal video creation and editing model that generates high-resolution video with audio from text, images, audio, and video inputs.

  • 🎥 Videos

Repositories

• 3d-shapes This repository contains the 3D shapes dataset, used in Kim, Hyunjik and Mnih, Andriy.
• dqn_zoo DQN Zoo is a collection of reference implementations of reinforcement learning agents developed at DeepMind b…

News

• AlphaGenome Atlas: Molecular predictions for 9 Billion human DNA variants AlphaGenome Atlas maps the molecular effects of 9 billion single-letter DNA variants across the human genome.

Papers

• CLIP-CLOP: CLIP-Guided Collage and Photomontage The unabated mystique of large-scale neural networks, such as the CLIP dual image-and-text encoder, populariz…
• Competition-Level Code Generation with AlphaCode Programming is a powerful and ubiquitous problem-solving tool. Developing systems that can assist programmers…
• Concrete Problems in AI Safety Rapid progress in machine learning and artificial intelligence (AI) has brought increasing attention to the p…
• Discovering faster matrix multiplication algorithms with reinforcement learning Formulated algorithm discovery as a single-player game and used AlphaZero-style RL to find novel matrix multi…
• Does This Gradient Spark Joy? Policy gradient computes a backward pass for every sample, even though the backward pass is expensive and mos…