See what it can do first

Before we get into the mechanics, let's look at what this hand can actually pull off. The official demos are all tasks that require finger coordination.

These aren't specially tuned demo modes. They're the natural outcome of packing enough independently force-controlled joints into a human-scale hand.

Where the ordinary robot gripper hits its ceiling

To understand the weight of this launch, you first need to see where today's humanoid robot hands are stuck. The common approach on the market is a position-controlled two-finger gripper: you give it a target position, and it opens or closes to reach it.

For a developer, a hand like that exposes exactly three actions: pick, place, push. Every application built on this kind of hand is just some combination of those three moves — done blind, eyes closed. The ceiling isn't in the software; it's at the end of the arm.

The deeper problem is in the transmission. To amplify a small motor's torque into something usable, this kind of hand typically uses a high reduction ratio of 100:1 to 200:1. The cost: before an external contact force can travel back to the motor, it's absorbed entirely by friction in the gears. The hand has no sense of what it touched or how much force it used — engineers can only mount a camera outside and guess what the fingers are doing.

This hand knows what it touched while it's touching it

The idea behind this new NEO hand is to have it read the outside world's response back in real time, at the same moment it's acting on it.

The key here is tendon drive: the motor doesn't sit in the joint — it's mounted in the forearm, pulling the finger joints remotely through cables that act like tendons. It's a bit like a puppet show: the motor is on a distant control stick, thin strings tug at the puppet's fingers, and the fingers themselves can stay very light.

There's also a quiet channel running in the background called proprioception. Because every joint is closed-loop controlled, the hand knows its own pose and how much force each part is applying without looking — the same way you can touch your two index fingertips together with your eyes closed.

How to split 25 joints for both dexterity and strength

Degrees of freedom refers to the number of independent directions a joint can move — roughly, one DOF corresponds to one point that can be controlled independently. 25 degrees of freedom means there are 25 spots on this hand that can each move independently. Beyond the count, how they're allocated matters more.

Enough force-controlled DOF at a human-hand scale adds up to both dexterity and strength. Here are the hand's hard numbers.

That much strength allows full-hand grasps, tool use, carrying things, opening doors, pushing a loaded cart, and precise pinching under load — all while keeping full dexterity. The ±0.2 mm positioning accuracy lets it work at the "small object" scale where most human labor actually happens.

The last half-millimeter, on the fingertip

Force control alone isn't enough — the information in that last half-millimeter at the fingertip has to come from skin. NEO's fingertips and surface are covered in high-resolution tactile skin that continuously measures three things.

The most direct use of this skin is catching slips in real time. The instant an object starts to slide, the shear-force channel picks it up, and the force-controlled joints immediately tighten or adjust the grip to stabilize it before the slide completes.

Vision alone can't do this, especially with objects that are transparent, fragile, deformable, or occluded. Official demos show a normal-force contact map, a pressure heatmap during a handshake, and pinching up a fragile paper crane without crushing it. This skin is designed together with the sensors inside it and the tendons behind it — it's a functional material, not a cosmetic layer.

What happens when it's hit with a hammer or caught in a drawer

A hand that learns by touch has to survive being touched, over and over — and it also has to be safe around people.

This hand's safety comes from its "compliance." The extremely low reduction ratio, combined with tendon drive and very low fingertip inertia, lets an external impact safely push the finger back and absorb the blow smoothly. In the official slow-motion clips, the fingers yield gracefully in every one of these situations.

Reliability is engineered into every subsystem: tendon routing, bearings, finger structure, cable routing, tactile integration, electronics, assembly. Individual parts and full finger assemblies have been tested through millions of cycles, drive units have been tested at extreme temperatures, and the wrist joint has been validated for over 2 million cycles under high load. The whole hand is sealed to IP68, uses food-safe materials, and can be washed by hand at a sink when it gets dirty.

Being able to manufacture it is the real moat

The last number — and the strategic point 1X is emphasizing — is production capacity.

First, how the hardware delivers all that force. The motors sit in the forearm, roughly where most of a human hand's grip strength comes from, pulling the fingers through 1X's proprietary tendons that run through the wrist. That's how the hand can stay light while producing serious force, and keep running continuously without overheating.

The whole hand is deeply integrated as part of the full NEO robot: in-house motors, custom electronics, embedded sensing, a proprietary tendon system, a compact transmission, and hand-specific firmware. From the tendon material down to the outermost soft polymer, the skin, and the tactile sensing stack — every hand is built end-to-end on 1X's own production line.

1X changed the approach: have the hand read the outside world's response back, at the same time it's acting. Push on one of its fingers, and it yields with the push while precisely reporting how much force you applied. 1X calls this "force transparency" (force and information flow both ways through the same channel). The method is "tendon drive" (motors sit in the forearm, pulling the finger joints remotely through cable "tendons"), paired with an unusually low reduction ratio.

It can genuinely do fine work that needs finger coordination: building LEGO, pinching coins out of a wallet, screwing in a lightbulb. So how exactly do force and information run both ways along "the same tendon"? One figure below makes it click.

Simple scenario: XiaoHu reaches out and pushes on one of NEO's fingers. The exact same push takes two completely different paths on the old gripper versus the new hand.

"A 5:1 reduction ratio" means nothing to most people, but it's exactly what decides whether this hand can "feel": the reduction ratio is the amplification factor of the gears between the motor and the joint — the higher it is, the more easily the force of pushing on the finger gets absorbed by gear friction. Here's how far apart the two approaches are:

All performance figures above come from 1X's own internal testing, with no third-party reproduction yet; the 10,000-unit figure is a planned 2026 capacity (a forecast), not something already achieved.