Gabriel Barth-Maron

Portrait of Gabriel Barth-Maron

Research Director · Google DeepMind · London

I build frontier multimodal generative models and the research systems behind them.

I currently co-lead Gemini Omni, a model family that brings Gemini's reasoning together with generative media. Previously, I co-led Veo across its first three generations.

Experience Google Scholar

Current work

Gemini Omni Create anything from any input Co-leading a model family that combines world understanding, multimodality, and conversational media editing. Explore Gemini Omni → Veo Advancing generative video Led strategy and execution across Veo 1, 2, and 3, from large-scale training to new model architectures. Explore Veo →

Earlier work

I helped build Gemini's multimodal foundations, led technical development for Gato, and pioneered distributed reinforcement learning at DeepMind. That work also produced the open-source systems Acme, Reverb, and Launchpad.

I hold a BA in Mathematical Economics and an ScM in Computer Science from Brown University.

news

May 19, 2026 Gemini Omni launched at Google I/O, bringing Gemini’s world understanding to video creation and conversational editing.
May 20, 2025 Veo 3 launched with major advances in video quality and native audio generation.
Jul 5, 2023 Transactions on Machine Learning Research (TMLR) awarded Gato its first Outstanding Certification (Best Paper Award). Read the full post here.
Jun 20, 2023 Gato was used as the backbone for RoboCat, a self-improving robotic agent that learns to perform tasks across different real-world robot arms.
Dec 9, 2022 Part of a panel discussion on Scaling + Models at the 5th Robot Learning Workshop at NeurIPS. You can find a recording here.

selected publications

  1. Gemini 2.5: Pushing the Frontier with Advanced Reasoning, Multimodality, Long Context, and Next Generation Agentic Capabilities
    Gemini Team
    Technical report 2025
    arXiv
  2. Gemini 1.5: Unlocking Multimodal Understanding Across Millions of Tokens of Context
    Gemini Team
    Technical report 2024
    arXiv
  3. Gemini: A Family of Highly Capable Multimodal Models
    Gemini Team
    Technical report 2023
    arXiv
  4. Distributed prioritized experience replay
    Dan Horgan, John Quan, David Budden, Gabriel Barth-Maron, Matteo Hessel, Hado Van Hasselt, and 1 more author
    arXiv preprint arXiv:1803.00933 2018
    arXiv
  5. Distributed distributional deterministic policy gradients
    Gabriel Barth-Maron, Matthew W Hoffman, David Budden, Will Dabney, Dan Horgan, Dhruva Tb, and 3 more authors
    arXiv preprint arXiv:1804.08617 2018
    arXiv
  6. A Generalist Agent
    Scott Reed, Konrad Zolna, Emilio Parisotto, Sergio Gomez Colmenarejo, Alexander Novikov, Gabriel Barth-Maron, and 5 more authors
    arXiv preprint arXiv:2205.06175 2022
    arXiv