Quick take
- Tesla Optimus remains in R&D: Musk told investors in January 2026 it's "not in usage in our factories in a material way," with only "several hundred units deployed, primarily for learning." A December 2025 Miami demo showing hand movements resembling a VR headset removal reignited teleoperation questions Tesla has never fully addressed.
- Figure 02 has the strongest verified industrial numbers of the four: an 11-month BMW Spartanburg pilot that assisted production of 30,000+ vehicles at greater than 99% placement accuracy. BMW has since expanded the partnership to Figure 03 for logistics work.
- 1X Neo is pre-selling into actual homes at $20,000 (or $499/month) — but by the company's own admission, roughly 30–40% of attempted tasks still route to a remote human operator inside the customer's home.
- Unitree G1 is the only one of the four already commercially shipping at real, listed prices since 2024 — but the base model's hands have zero degrees of freedom, making it a mobility platform, not a dexterous labor substitute.
- The company actually shipping product at scale is also the most cautious in public: Unitree's Wang Xingxing said humanoid robots' "ChatGPT moment" could be "up to 10 years away" — a sharp contrast to Musk's and Figure's public predictions.
Four companies, four very different definitions of "ready." Tesla Optimus, Figure 02, 1X Neo, and Unitree G1 are all being marketed as the humanoid robot that finally gets out of the lab — but as of September 2026, they're at completely different stages of an honesty problem the industry hasn't solved: the gap between what a robot does in a demo reel and what it actually does, reliably, without a human quietly steering it.
Tesla Optimus: still R&D, despite the framing
Musk has called Optimus "our biggest product, not just Tesla's biggest product ever, but probably the biggest product ever," and repeated that framing on Tesla's April 2026 earnings call. The company's own numbers tell a more modest story: at the January 2026 Q4 earnings call, Musk admitted Optimus "is still in the R&D phase" and "not in usage in our factories in a material way," with only "several hundred units deployed, primarily for learning, not productive tasks." Tesla has publicly acknowledged that no mature supply chain exists yet for Optimus's components — they have to be developed from scratch.
Production has repeatedly missed its own targets: 5,000 units planned for 2025 didn't happen, and the 2026 goal of 50,000 is badly behind schedule. On the August 2026 Q2 call, Musk said Optimus 3 production was starting "soon" at Fremont, with an eventual Optimus 4 planned for Austin at "an order of magnitude more" scale. A December 7, 2025 demo in Miami showed Optimus falling, with hand movements some observers compared to someone removing a VR headset — reigniting a teleoperation debate Tesla has never fully put to rest, after its October 2024 "We, Robot" event's bartending and crowd-interaction demos were later confirmed to be human-teleoperated.
Investor confidence has followed the pattern: prediction-market odds of a commercial Optimus launch in 2026 fell from a 33% peak to just 9% by late August 2026, according to Benzinga's tracking of Polymarket data.
Figure 02: the strongest real numbers
Figure AI's Figure 02 has the most credible deployment data of the four, from an 11-month pilot at BMW's Spartanburg, South Carolina plant, where the robots performed sheet-metal weld-fixture insertions for X3 production. Figure's own release states the robots contributed to production of over 30,000 vehicles, logged 1.2 million steps, traveled 200+ miles, and hit greater than 99% placement accuracy per shift, consistently meeting an 84-second target cycle time.
Following that pilot's success, BMW expanded the partnership: Figure 03 is now being deployed for a new logistics task, sequencing unsorted components into trolleys ahead of autonomous delivery to the assembly line. CEO Brett Adcock has said the "home is coming" for humanoid robots, predicting useful home deployment is "single-digit years away," and that Figure expects to manufacture roughly 100,000 units within four years. Unlike the other three, Figure 02/03 has no published consumer or third-party commercial price — it remains deployed only through direct industrial partnerships like BMW's.
1X Neo: honest about the human inside it
1X Neo is the furthest along toward an actual home product, and also the most transparent about its limitations. Pre-orders opened in 2026 at $20,000 with a $200 deposit for priority delivery, or a $499/month subscription. The robot weighs 66 pounds, can lift over 150 pounds, carries up to 55 pounds, and runs at 22 decibels — quieter than a refrigerator.
The central fact reporters keep returning to: 1X has stated Neo completes roughly 60–70% of attempted tasks autonomously, with the rest routed to a remote human operator working inside the customer's home network. An investigative piece from Live in the Future, headlined "1X Sold 10,000 Robots in 5 Days. Each One Needs a Human Hiding Inside It," frames this as labor arbitrage dressed as autonomy — a claim not independently corroborated elsewhere in this research, but consistent with 1X's own public statements about the split. The company's stated safeguards include blurring operators' view of faces, user-configurable no-go zones, and requiring approval before each remote session connects.
Unitree G1: shipping, but with zero-DOF hands
Unitree G1 is the only one of the four already commercially shipping at real, listed prices — it launched in 2024 and has been selling steadily since. The base G1 dropped from its original $16,000 launch price to $13,500 by late 2025 (backordered as of mid-2026); US dealer pricing with domestic support runs $17,990. Research-grade "EDU" configurations with far more dexterity run $43,900–$73,900.
The catch is dexterity: the base G1 has 23 total degrees of freedom, but its stock "dummy hands" have zero degrees of freedom — no functional fingers or wrist articulation. It's a mobility and research platform out of the box, not a dexterous labor substitute, unless you buy into the far more expensive EDU tier. Battery life is roughly 2 hours of normal use, dropping to about 1 hour under high-load dynamic motion like running or jumping.
Optimus
Several hundred units, "learning" only. Not material factory use as of early 2026. Repeated teleoperation controversies.
Figure 02 / 03
30,000+ BMW vehicles assisted, >99% placement accuracy. Not sold commercially — deployed only via direct industrial partnerships.
Neo
66 lbs, lifts 150+ lbs. Company states ~60-70% task autonomy; the rest routes to a remote human operator.
G1
23 DOF, but stock hands have zero DOF. ~2 hrs battery, ~1 hr under high load. EDU tier adds real dexterity at much higher cost.
The person actually shipping product is the most cautious
Unitree shipped 5,500 humanoid units in 2025 and is targeting 20,000 for 2026 — roughly a fourfold increase — per CEO Wang Xingxing. Unitree completed an IPO in 2026, making the 36-year-old Wang, per Forbes, "China's first humanoid robot billionaire." Despite running the company with the clearest commercial traction of the four, Wang was notably more restrained than Musk or Adcock when asked about the industry's real breakthrough moment.
Globally, Chinese manufacturers accounted for roughly 97% of an estimated 19,000 humanoid units shipped in the first half of 2026 — up 272% year over year — with Morgan Stanley forecasting China alone could hit 50,000 units for the full year, up from 12,000 in 2025.
No confirmed safety incidents — but no humanoid-specific rules yet either
This research did not find a documented, named injury or safety incident tied to Optimus, Figure 02/03, Neo, or G1 specifically. General industrial-robot injury data exists — OSHA attributes more than 60% of robot workplace accidents to unexpected activation — and regulation is catching up in pieces: the EU AI Act's high-risk classification took effect August 2, 2026, though OSHA had no humanoid-specific guidance as of mid-2026.
| Metric | Figure |
|---|---|
| Optimus units deployed (learning only) | "Several hundred" |
| Optimus 2026 production target | 50,000 (badly behind) |
| Figure 02 at BMW | 30,000+ vehicles, >99% accuracy |
| 1X Neo price | $20,000 or $499/mo |
| 1X Neo company-stated autonomy | ~60-70% of tasks |
| Unitree G1 base price | $13,500 (0-DOF hands) |
| Unitree 2025 shipments / 2026 target | 5,500 / 20,000 |
| Global H1 2026 humanoid shipments | ~19,000 (+272% YoY) |
Common questions
Is Tesla Optimus actually working in Tesla's factories?
Not materially, as of early 2026 — Musk himself said it's "not in usage in our factories in a material way," with only a few hundred units deployed for learning purposes.
Which humanoid robot has the best real-world results?
Figure 02, based on its BMW Spartanburg pilot: 30,000+ vehicles assisted at over 99% placement accuracy. It's not sold commercially, though — only deployed through direct industrial partnerships.
Does 1X Neo actually work on its own?
Partially. 1X states Neo completes roughly 60-70% of attempted tasks autonomously; the rest are handled by a remote human operator inside the customer's home.
Can I buy a humanoid robot today?
Unitree G1 is the only one of the four already commercially available, from $13,500 — but its base "dummy hands" have zero degrees of freedom, limiting real dexterity out of the box.
Has any of these robots been in a documented safety incident?
No named, confirmed injury incident tied to any of these four specific robots was found in this research.
Strip away the marketing and the four robots sort into distinct, honest categories: Figure 02 has the best verified industrial results but isn't for sale. Unitree G1 is genuinely for sale and scaling fast, but its base model's hands can't do much. 1X Neo is pre-selling into real homes while openly admitting a third of its work still needs a hidden human. And Optimus, despite carrying the loudest marketing of the four, remains the least verified — "several hundred units" deployed for "learning," with a documented history of demos later revealed as teleoperated. The person running the company that's actually shipping product at scale says the real breakthrough is still a decade out. That's probably the most honest number in this entire category right now.
Reporting note. This report was independently researched and written by DopamineKart's editorial desk. No company named above reviewed, sponsored, or paid for this coverage. Optimus's degrees of freedom and battery-life specifications were not found with confirmed 2026 sourcing and are omitted rather than estimated. No consumer or third-party commercial price for Figure 02/03 has been published. The "10,000 Neo units in 5 days" claim comes from a single outlet (Live in the Future) and was not independently corroborated elsewhere in this research. Photography: Wikimedia Commons only, credited per image — no manufacturer-owned photography appears on this page.
Corrections: email us