Google DeepMind releases Gemini Robotics 2.5 for real-world tasks
Google DeepMind: Gemini Robotics 2.5 accelerates real-world general dexterity
Our most capable vision-language-action model enhances robot dexterity, reliability, and safety
Google DeepMind introduced Gemini Robotics 2.5, a vision-language-action model that improves robot performance on real-world tasks. The model increases task success rates by 30% to 100% compared to previous versions while completing tasks 20% to 30% faster across over 200 distinct tasks. In factory testing, robots achieved a 96% success rate at repetitive tasks like picking and packing automotive parts.
- Google DeepMind ↗: The model is designed to work with multiple robot types and is rolling out to robotics partners to help robots move from laboratory environments into unstructured, real-world settings.
- Ars Technica ↗: DeepMind also released an open source safety benchmark called ASIMOV-Jailbreak containing 1,500 prompts designed to test whether robotic AI can safely refuse unsafe requests.
The improvements enable robots to assist people with everyday work in real-world environments. The safety benchmark addresses the need for standardized safety testing as robots begin working outside controlled laboratory environments.
Google DeepMind has introduced Gemini Robotics 2.5, a vision-language-action model designed to enable robots to learn and perform a wider range of tasks more quickly in real-world settings. The model improves robot dexterity, reliability, and safety by increasing task success rates by 30% to 100% compared to previous versions while completing tasks 20% to 30% faster across over 200 distinct tasks including tool use and handling delicate objects. In a real-world test, the model helped a humanoid robot pick and pack automotive parts with a 96% success rate on production lines. Gemini Robotics 2.5 is designed to work with multiple robot types and is rolling out to robotics partners to help robots move from laboratory environments into unstructured, real-world settings where they can assist people with everyday work.
One short email a day with the stories that matter. Free. Unsubscribe with one click.
Want to go deeper than the news? Explore live, cohort-based AI courses taught by practitioners.
Browse AI courses on Maven
Google DeepMind releases new advanced robotics AI model and next-generation gene prediction model
Alphabet Inc.’s artificial intelligence research lab, Google DeepMind, Tuesday debuted the next generation of Gemini Robotics, the company’s most capable vision-language-action model to date for creating an “AI brain” for robotics. Separately, DeepMind announced AlphaGenome 2 today, a next-generation AI model that can predict how DNA sequences and their variations affect human cell types, creating […] The post Google DeepMind releases new advanced robotics AI model and next-generation gene
Also reported by MarkTechPost
Can AI agents program industrial robots without training data? Berkeley says yes
The new RoboSimStudio is an open-source project that enables agentic AI coding of robotic simulation environments. | Credit: UC Berkeley UC Berkeley and NVIDIA have released open-source tools for a new approach: enabling teams of AI coding agents to write, simulate, and refine readable robot programs. Since the early days of industrial robots, programming, tuning, and troubleshooting robotic workcells has taken weeks to months. New research coming out of UC Berkeley has demonstrated that new age
Boston Dynamics adds wheels to Atlas humanoid to test whole-body control
Bridging the divide between deterministic physics-based control and policy-driven adaptation, Boston Dynamics has deployed a hybrid locomotion architecture on a wheeled variant of its Atlas humanoid. The “experiment” combined classical model-based motor control with whole-body reinforcement learning to maintain dynamic stability across unstructured, real-world terrain. In the video, Andy Wu, controls engineer for Atlas at Boston Dynamics, highlights some recent experiments that he
Practice Makes Precise: At CoRL, NVIDIA and Partners Reveal How Robots Rehearse Delicate Work in Simulation
Simulation, which has long been the proving ground for robots learning to walk, roll and hop, is now helping researchers teach robots tasks that require a finer touch: aligning gears, connecting cables and assembling small components. The shift from locomotion to manipulation is taking center stage at this week’s Conference on Robot Learning (CoRL) in Austin, Texas, alongside the rise of agentic robotics — a new paradigm that integrates frontier agents to create interpretable code for physical
Memories.ai builds standardized egocentric video pipeline
Memories.ai models label every clip: Memories-H0 for hand pose, EgoAction for action segments. | Credit: Memories.ai Memories.ai is building a data pipeline for robotics companies that combines standardized video collection with AI-powered annotation to turn everyday egocentric footage into training data for robots. The San Francisco-based company is developing software and AI to help robots understand the physical world. It builds video-understanding models and data infrastructure, and it colle
Dynamic Creatures partners with Sixth Wall Studio to bring Hasbro characters to life with robots
On the left, Roberta Thomson, the CEO of Sixth Wall Studio, and on the right, Marc Theerman, co-founder and CEO of Dynamic Creatures. | Source: Dynamic Creatures Dynamic Creatures today said it has partnered with Sixth Wall Studio, an AI studio launched by Hasbro in June 2026. The collaborators plan to bring Hasbro characters to life through interactive robots. The company will combine its mobile, expressive robots with Sixth Wall‘s character authoring and alignment. The goal is to create