GENE-26.5 brings dexterous AI robotics into focus

Genesis AI has introduced GENE-26.5, a robotics foundation model designed to enable robots with human-level physical manipulation capabilities. The system integrates an AI-driven 'brain' capable of handling complex, long-horizon tasks such as cooking, preparing smoothies, lab automation, and solving a Rubik’s Cube. The company demonstrated its capabilities in a video, highlighting tasks like wire harnessing and multi-object grasping. A key component of GENE-26.5 is Genesis AI’s dexterous robotic hand, paired with a tactile-sensing glove that mimics human hand movements. This 1:1:1 mapping between the glove, human hand, and robotic hand allows for direct skill transfer from human demonstrations to robotic applications. The glove is claimed to be 100 times cheaper than conventional options and achieves up to five times greater data collection efficiency in internal tests. Genesis AI plans to deploy the glove with partners in real-world environments to build a human skill library, addressing a major bottleneck in robotics training. The company also leverages simulation technology to narrow the gap between virtual and real-world training, aiming to accelerate model development and evaluation. Co-Founder and CEO Zhou Xian stated that GENE-26.5 represents the industry’s most advanced robotic brain and hand, enabling reliable, scalable manipulation tasks previously limited to human hands. Co-Founder and President Theophile Gervet emphasized the company’s full-stack innovation approach, from AI to hardware, as a competitive advantage in harnessing unprecedented data for foundation models. The announcement underscores Genesis AI’s focus on solving robotics challenges holistically, combining AI, hardware, and data collection to push the boundaries of robotic dexterity. The company’s goal is to make robots capable of performing complex, real-world tasks with human-like precision and efficiency.