August 2, 2026
On July 9, 2026, 1X announced a redesigned hand for its NEO home robot, and the spec sheet is unusually specific for a company about to ship into actual living rooms rather than run a lab demo. Forbes reported the numbers: 25 total degrees of freedom, with 22 actuated across the fingers and palm and 3 more in the wrist, tendon-driven by motors mounted up in the forearm rather than crammed into the hand itself. Add tactile sensing that reads normal force, contact location, and shear, real-time slip detection, up to 45 N of fingertip flexion force, positioning precision of plus or minus 0.2mm, IP68 waterproofing, and food-safe materials. 1X says it can build roughly 10,000 of these hands in-house during 2026. That's a lot of engineering detail to put in a press release, and it tells you something: 1X thinks the hand is the part of NEO that skeptics need convincing on. They're right to think that. Walking on flat floors and turning through doorways is a solved-enough problem that most serious humanoid companies have working demos of it. Manipulating the actual physical clutter of a household is not solved anywhere, by anyone, at consumer prices.
Walking is a mostly-known problem because the inputs are predictable: gravity, a flat-ish floor, a body with fixed dimensions. You can model it, simulate it, and tune a controller against a narrow band of real-world variation. A home is nothing like that. Every dish is a different shape and weight. A wine glass and a cast-iron skillet need completely different grip force and finger spread, and the robot doesn't get to know which one it's about to touch until its sensors tell it in the moment. That's why the tactile sensing matters more than the degrees-of-freedom count. Twenty-two actuated joints in the hand is impressive, but a hand with great range of motion and no sense of touch will crush an egg or drop a mug, because it's working blind. Slip detection -- knowing in real time that an object is sliding out of grip before it hits the floor -- is the difference between a robot that can be trusted around a toddler's toys or a dog's water bowl and one that needs a clean, empty test bench. The 45 N of fingertip force and the 0.2mm precision are the specs that make headlines, but slip detection and shear sensing are the specs that determine whether NEO can operate in a kitchen that hasn't been staged for a video.
A staged demo proves nothing anymore -- every humanoid company has clean, well-lit clips of a robot placing a single item on a counter. The test that would actually convince me is something boring and unscripted: loading a dishwasher with a random mixed set of a household's actual dishes -- a chipped ceramic bowl, a stemmed glass, a nonstick pan with a loose handle -- sorted from a sink where they're stacked and touching, not laid out with gaps for the camera. Add pulling laundry apart when a sock is balled up inside a fitted sheet, or picking a dropped set of keys off a rug without leaving a dent in a memory-foam surface. None of that requires elegance. It requires the hand to handle ambiguity it wasn't specifically trained on, which is exactly what tactile sensing and slip detection are supposed to buy you. Until 1X or a reviewer publishes footage of NEO doing something like that, unedited, in a home that wasn't tidied for the cameras, the spec sheet is a strong claim, not a demonstrated result. Specs describe what the hardware can theoretically do; they don't describe what the perception and control software actually does with that hardware when the countertop is a mess.
SlashGear corroborated the pricing: $20,000 outright, which includes a three-year warranty, priority delivery, and premium support, or $499 a month with a refundable $200 deposit, and daily electricity cost under $1. As a subscription, $499 a month is priced like a luxury car lease, and for a household that can genuinely offload laundry-folding, dish-loading, and countertop-clearing to it, that's a real value proposition, not a novelty purchase. But value only shows up if the robot reliably does those tasks unsupervised, and 'reliably' is exactly the part these specs don't settle. My honest read: this is viable for a narrow slice of early-adopter households this year -- people with disposable income, tolerance for a first-generation product, and tasks simple enough to fall inside whatever NEO can actually do at launch. It's not viable yet as a mainstream appliance purchase, the way a dishwasher or a robot vacuum is. The gap between 'has 45 N of fingertip force and slip detection' and 'my in-laws don't need to watch it' is still a software and reliability gap, not a hardware spec gap, and 1X hasn't published the number that would close that gap: task success rate in unstructured homes over weeks of continuous use, not degrees of freedom.
I'd update fast on a few concrete things: independent, unedited footage of NEO handling a genuinely cluttered kitchen or laundry pile in a home that wasn't prepped for it; a published failure rate or incident count from early home deliveries later in 2026, not just a highlight reel; and evidence that the hand's tactile sensing generalizes to objects it's never seen, rather than a curated set 1X trained specifically on. Until then, the honest position is that 1X has published the most detailed and credible hand spec of any humanoid company aiming at real homes -- which is genuinely notable -- while the hardest part of the problem, generalizing gracefully to a home nobody optimized for the robot, remains unproven. For more of our ongoing read on where humanoid robotics actually stands versus the hype, see our recent coverage of the humanoid robot policy fight and why the robotics news that isn't about humanoids is often the more telling story.
Sources