On March 12 2025, Google DeepMind unveiled Gemini Robotics, a vision‑language‑action model built on the Gemini 2.0 large language model, together with a companion version named Gemini Robotics‑ER, which stands for embodied reasoning. The announcement was made in collaboration with Apptronik and represents a notable advance in applying artificial intelligence to robotic systems. Gemini Robotics is engineered to interpret unfamiliar situations, a trait that makes it adaptable for a wide range of robotic tasks. By adapting the Gemini 2.0 language model for robotics, the developers intend to improve how machines perceive and act within their surroundings.
The embodied reasoning component of Gemini Robotics‑ER adds a layer of contextual understanding, aiming to enable more sophisticated and nuanced interactions between robots and their environments. Access to the models is currently restricted.
Google DeepMind has provided early access to a select group of trusted testers, including Agile Robots, Agility Robotics, Boston Dynamics, and Enchanted Tools. This controlled rollout allows the team to gather feedback, refine performance, and ensure the technology meets high standards of quality and reliability before broader distribution. The potential applications span multiple sectors.
In manufacturing and logistics, the models could help robots handle variable workflows and unexpected obstacles. In healthcare and education, they might support machines that need to interpret complex human cues and adapt to dynamic settings.
The developers suggest that, as the technology matures, it could lead to more autonomous and precise robotic systems across these fields. On June 24 2025, Google DeepMind released Gemini Robotics On‑Device, a variant optimized to run directly on robotic hardware. By enabling local execution, the model reduces reliance on constant cloud connectivity, which can improve responsiveness and allow operation in environments with limited or intermittent network access.
This on‑device version is expected to broaden adoption and simplify integration into various robotic platforms. The ongoing partnership between Google DeepMind, Apptronik, and the selected tester companies is intended to drive iterative improvements.
Insights gathered from real‑world testing will inform future updates, potentially expanding the models’ capabilities and addressing any limitations identified during the trial phase. Looking forward, the trajectory of Gemini Robotics and its related versions will likely involve further refinements, broader availability, and new use‑case explorations. As the technology becomes more widespread, stakeholders anticipate advances in robotic autonomy and precision, accompanied by the need to consider both benefits and challenges associated with deploying advanced AI‑driven machines in everyday contexts.
{“@context”: “https://schema.org”, “@type”: “NewsArticle”, “headline”: “Google DeepMind Launches Advanced Gemini Robotics AI Model Successfully”, “description”: “On March 12 2025, Google DeepMind unveiled Gemini Robotics, a vision‑language‑action model built on the Gemini 2.0 large language model, together with a compani”, “datePublished”: “2026-06-22T09:30:00+00:00”, “author”: {“@type”: “Person”, “name”: “Li Wei”}, “publisher”: {“@type”: “Organization”, “name”: “InfoPulse Today”, “url”: “https://newsfeed.infopulsetoday.com/”}, “image”: “https://infopulsetoday.com/wp-content/uploads/sites/4/rehomed/s25/2026/05/google-deepmind-gemini-robotics-apptronik.png”}
