Genesis AI Unveils GENE-26.5 Model for Enhanced Robotic Dexterity

The GENE-26.5 robot brain enables dexterous, two-handed cooking. Source: Genesis AI

After years of development, dexterous robotic manipulation is advancing significantly, driven by breakthroughs in artificial intelligence. Genesis AI has unveiled GENE-26.5, a model it claims provides robots with "human-level physical manipulation capabilities."

The San Carlos, Calif.-based company also announced a system designed to "unlock unlimited amounts of data" for training GENE-26.5 at scale. This system integrates two proprietary components: a novel data engine and a human-scale robotic hand that facilitates direct skill transfer from humans to robots.

"Together, these innovations address the fundamental data bottleneck that has limited robotics foundation models, paving the way for a new generation of highly productive general-purpose robots," stated Genesis AI.

"The brain and the hand are the two most critical and complex elements in robotics. Today, we are introducing the industry's most advanced versions of both," said Zhou Xian, co-founder and CEO of Genesis AI. "For the first time, we are enabling robots to perform tasks once exclusive to human hands, doing so reliably and at scale."

The startup emerged from stealth last year with $105 million in funding. Genesis AI describes itself as a "global full-stack robotics company building general-purpose robots with human-level intelligence and capabilities."

GENE-26.5 designed for complex robot tasks

"GENE-26.5 is an AI foundation model purpose-built for robotics, engineered to process massive datasets and adapt to diverse environments," explained Genesis AI. It aims to empower robots to execute complex, long-horizon tasks with human-like dexterity. The company positions GENE-26.5 as a significant step toward a future where robots can be rapidly deployed and adapt to new environments and unfamiliar tasks.

To showcase GENE-26.5's capabilities, Genesis AI released a video demonstrating its system's fluid, human-like dexterity and precise, coordinated hand movements across several complex tasks:

• Cooking a multi-step meal, including chopping tomatoes, cracking an egg one-handed, and coordinating both hands
• Preparing a smoothie, involving ingredient handling, pouring, blending, and mid-air serving with coordinated hand control
• Conducting high-precision lab experiments with delicate instrumentation, such as pipetting, liquid transfer, and mid-air manipulation
• Wire harnessing: arranging and securing wires into organized bundles—a task Genesis AI notes is among the most challenging in electronics and electrical engineering
• Solving a Rubik's Cube using continuous in-air manipulation that requires coordinated, high-speed reasoning and precise wrist control
• Single-handed multi-object grasping, simultaneously managing four objects of varying sizes and sorting them into designated bins
• Playing the piano at a human level, performing an ultra-fast, highly complex composition

Genesis AI stated that GENE-26.5 has demonstrated its ability to equip robots with a broad spectrum of complex skills and a level of physical manipulation previously unattainable.

"General-purpose robotics has the potential to reshape the global economy while opening an entirely new chapter for AI," said Eric Schmidt, former CEO of Google and an investor in Genesis AI. "Genesis is introducing a paradigm shift in robotics. This marks an important milestone for their team and the broader robotics industry."

Robotic gripper mirrors the human hand

"The embodiment gap—the difference between human and robotic form—has historically severely limited robot performance and their ability to learn from human data," Genesis AI noted.

The company claims to have addressed these limitations with its proprietary hardware, which closely mimics the human hand. The end effector mirrors the human hand in both form and function and pairs with a data-collection glove equipped with tactile-sensing electronic skin.

"When worn by a human, the glove enables a precise 1:1:1 mapping between the glove itself, the human hand, and the robotic hand," according to Genesis AI. "This allows humans to seamlessly provide GENE-26.5 with high-quality data at scale, which translates directly into robotic skills. The result is natural, human-like movement that generalizes reliably across complex physical tasks, use cases, and environments."

Editor’s note: At the 2026 Robotics Summit & Expo this month in Boston, keynotes, sessions, and exhibits will focus on embodied and physical AI. Register now to attend.

Low-cost hardware enables scalable data collection

Genesis AI also stated that its data-collection glove is 100 times cheaper than typical hardware and has demonstrated up to five times greater data-collection efficiency than traditional teleoperation methods in internal testing. This approach is said to enable affordable, continuous, large-scale robotics training.

The company is engaging with partners to deploy the glove in real-world work environments. By simply wearing the glove during routine tasks, everyday activities can become sources of new categories of training data, contributing to what Genesis AI calls the world's largest human skill library.

Furthermore, Genesis AI's data engine leverages egocentric video data from humans wearing cameras to capture real-world interactions, alongside vast amounts of human-centric internet videos. The company asserts this methodology will allow its foundation model to learn more efficiently and enable robots to perform increasingly complex tasks.

"At Genesis, we believe success in robotics requires excellence at every level," said Theophile Gervet, co-founder and president of Genesis AI. "That's why we're obsessed with innovating across the full stack, from AI to hardware. By controlling every layer, we can build a cohesive system and solve the problem holistically. Our approach provides a significant competitive advantage by harnessing unprecedented volumes of data, which ultimately defines what foundation models can achieve."

GENE demonstrates automated laboratory pipetting. Source: Genesis AI

GENE uses simulation to accelerate training

"Robot development has traditionally been one of the most manual processes in engineering," said Genesis AI. The company has developed a simulation system to narrow the sim-to-real gap—the difference between virtual and physical worlds. Its rendering engine and hyper-realistic physics are designed to create accurate and reliable representations of real-world conditions, allowing robots to perform tasks as they would in reality.

This level of simulation is claimed to accelerate the development, testing, and refinement of its robotic AI stack. Teams can train and evaluate models far more rapidly than with traditional physical testing, which is often slow, costly, and difficult to scale.

"Genesis is changing the trajectory of robotics, bringing us closer to AI that can operate effectively in the real world," said Vinod Khosla, founder of Khosla Ventures and a Genesis AI investor. "Genesis's groundbreaking foundation model, powered by its human-centric data engine and its first-of-its-kind simulation, is poised to dramatically increase development speeds and enable instant deployment with commercial customers."

Genesis AI announced it will soon reveal its first general-purpose robot based on the technology unveiled today. The company is backed by Eclipse, Bpifrance, and HSG, alongside technology visionaries such as Eric Schmidt and Xavier Niel, and AI pioneers Daniela Rus and Vladlen Koltun.

Genesis AI’s foundation model enables its robot to solve a Rubik’s Cube. Source: Genesis AI