Humanoid Robot Specs Compared: 2026 Mid-Year Data Breakdown — Price, Payload, DOF & Real-World Value

Why Raw Specs Don’t Tell the Whole Story — But the Math Does

With more than a dozen commercially active or near-commercial humanoid robots on the market as of June 2026, procurement teams and integration partners are no longer asking if humanoids work — they’re asking which one is worth the investment. To answer that rigorously, we’ve moved beyond spec sheets and calculated derived metrics: price per degree of freedom, payload-to-weight ratios where data permits, and total cost of ownership signals. What emerges is a surprisingly differentiated landscape where the right robot depends almost entirely on deployment context.

Price Per Degree of Freedom: The Efficiency Metric Nobody Publishes

Degrees of freedom (DOF) is a proxy for mechanical dexterity — but only when normalized against cost does it reveal real value. Here’s what the current data shows across publicly priced platforms:

Unitree Introducing | Unitree R1 Intelligent Companion Price from $5900 — Unitree Robotics

  • Unitree G1 (43 DOF configuration): ~$313–$372 per DOF at $13,500–$16,000. This is the lowest price-per-DOF of any commercially available full humanoid in 2026 by a significant margin.
  • Tesla Optimus Gen 2 (28 DOF): ~$714–$1,071 per DOF at a target price of $20,000–$30,000 — though it remains unavailable at scale as of this writing.
  • Figure 03 (44 DOF full body): target below $20,000, implying under $455 per DOF if pricing holds — competitive with Unitree if Figure AI hits its consumer target. Current availability: not yet shipping.
  • Fourier GR-2 (53 DOF): ~$2,830–$3,207 per DOF at $150,000–$170,000. Premium pricing reflects enterprise-grade build quality and a 40 kg payload specification that no comparably priced platform matches.
  • XPENG IRON (82 active DOF / 200 total DOF): ~$1,829 per active DOF at an estimated $150,000. However, if the 200-total-DOF figure is used, the cost drops dramatically — though industry convention typically counts actuated joints.
  • AgiBot A2 Max (67 DOF): enterprise pricing undisclosed, making direct comparison impossible. The A2 Max holds the highest disclosed DOF count among non-XPENG platforms.

The DOF metric has limits. A robot with 43 DOF but imprecise actuation delivers less real-world value than one with 28 DOF and sub-millimeter repeatability. Boston Dynamics Atlas, for instance, discloses neither DOF nor payload publicly — yet commands enterprise contract pricing and is already deployed at Hyundai’s Metaplant in Georgia as of February 2026. Capability reputation carries its own premium that spreadsheets struggle to capture.

Payload-to-Robot Capability: Where the Numbers Actually Matter

For logistics and manufacturing buyers, payload is the defining specification. Here’s the ranked breakdown by disclosed carry capacity:

  • Fourier GR-1/GR-2: 40 kg — the highest disclosed payload of any platform with public pricing. At $150,000–$170,000, this targets heavy-duty manufacturing and warehouse applications where human workers handle bulky components.
  • Unitree H1/H1-2: 30 kg carry at $90,000–$150,000. Strong payload-per-dollar ratio for a platform that is currently available and shipping, with commercial distribution now expanding into Japan.
  • Apptronik Apollo: 25 kg — deployed at Mercedes-Benz facilities and backed by near-$1B in funding. Enterprise contract pricing only, but real-world validation at automotive scale is established.
  • Figure 03: 20 kg total carry — with 16 DOF per hand, the dexterity-to-payload balance is notable. BMW Spartanburg deployments, scaled up from February 2025, validate this specification in production conditions.
  • Agility Robotics Digit: 16 kg carry — available via RaaS at approximately $250,000 purchase equivalent. Digit’s $300M+ in pre-orders and $2.5B SPAC IPO signal that the market accepts this payload tier at premium subscription pricing for warehouse automation.
  • UBTECH Walker S1/S2: 8 kg per arm — with 5,000 units sold and 3,800+ pre-orders reported, Walker demonstrates that mid-payload robots can achieve commercial scale. The Siemens enterprise partnership adds integration credibility.
  • Tesla Optimus Gen 2: 9 kg per hand — a per-hand figure rather than total carry, making direct comparison ambiguous. With 11 DOF per hand and active Gigafactory Shanghai deployment since January 2025, dexterity at moderate payload appears to be the design priority.
  • 1X NEO: ~5 kg per arm at $20,000 or $499/month — the only platform explicitly designed for home use now reaching consumers, where light-payload dexterous tasks (folding laundry, carrying groceries) define success, not industrial lifting.
  • Unitree G1: 3 kg per arm — the lowest payload among full humanoids reviewed. At $13,500–$16,000, this is a research and light-task platform, not a logistics workhorse. Its commercial North America release is confirmed for June 30, 2026.

Use-Case Value Map: Matching Specs to Deployment Reality

Heavy Manufacturing & Automotive Assembly

For automotive-grade environments, the Fourier GR-2’s 40 kg payload and 53 DOF at $150,000–$170,000 represents the strongest specification-per-dollar in heavy-duty categories. The Unitree H1/H1-2 at 30 kg and up to $150,000 is a credible alternative, particularly given active commercial availability. Apptronik Apollo’s Mercedes-Benz deployment and Figure 03’s BMW Spartanburg scale-up demonstrate that 20–25 kg platforms handle real automotive tasks — but buyers should note that both operate on enterprise contracts without transparent per-unit pricing.

Warehouse & Logistics Automation

Agility Digit’s RaaS model (~$250,000 equivalent) with 16 kg payload and $300M+ in pre-orders positions it as the logistics-specific platform with the deepest commercial commitment. UBTECH Walker’s 5,000 units sold metric — the highest confirmed sales volume of any platform in this analysis — signals operational maturity. For buyers prioritizing proven fleet scale over maximum payload, Walker S1/S2 deserves serious evaluation despite its enterprise-only pricing opacity.

Research, Education & Prototyping

The Unitree G1 at $13,500 base price with up to 43 DOF is unambiguously the best value for research institutions and innovation labs — evidenced by its CES 2026 commercial launch targeting academic labs and training centers. The Unitree R1 at from $4,900 is worth monitoring for constrained budgets, though specification data remains limited. The 1X NEO at $499/month offers a subscription alternative for teams that need consumer-environment testing without capital commitment.

Consumer & Home Deployment

Only 1X NEO has confirmed consumer home deliveries in 2026, making it the sole platform with real data in this segment. Its teleoperation-first learning model is pragmatic given where AI task generalization currently stands. Tesla Optimus Gen 2’s $20,000–$30,000 target price would be compelling at scale, but with Gen 3 production ramp announcement confirmed for July 15, 2026, procurement teams should wait for updated specifications before committing to Gen 2 assumptions.

The Specs That Are Missing — And Why That Matters

A striking pattern across this dataset: the robots with the strongest market momentum often disclose the least. Boston Dynamics Atlas — deployed at Hyundai Metaplant since February 2026 with a confirmed commercial expansion on July 8, 2026 — discloses neither DOF nor payload publicly. Apptronik Apollo, fresh from a near-$1B funding round and active Mercedes-Benz deployment, lists only payload. This opacity is a deliberate enterprise positioning strategy: when your robot is already on an automotive factory floor, you negotiate contracts, not spec sheets.

For buyers, this means the most capable platforms in production today are the hardest to evaluate on paper. The practical implication: enterprise procurement timelines for Atlas, Apollo, and Digit should account for extended evaluation periods and pilot deployment phases that smaller-budget research buyers simply don’t face when ordering a Unitree G1 online.

The XPENG IRON Outlier: 200 DOF and What It Actually Means

XPENG IRON’s specification of 82 active DOF and 200 total DOF at an estimated $150,000 is the most architecturally ambitious figure in this dataset. Mass production began in January 2026 at the Guangzhou facility, leveraging EV manufacturing infrastructure. The gap between 82 active and 200 total joints suggests a high proportion of passive or compliant elements — common in bio-inspired designs intended to absorb impact rather than generate motion. Until XPENG publishes task performance data from its internal factory deployments, the 200 DOF headline should be read as an architectural philosophy, not a direct capability comparison against 44 DOF platforms with proven automotive deployment histories.

Bottom Line: The 2026 Value Rankings by Segment

  • Best price-per-DOF (available now): Unitree G1 at ~$313–$372/DOF
  • Best payload-per-dollar (available now): Unitree H1/H1-2 at 30 kg up to $150,000
  • Best enterprise payload specification: Fourier GR-2 at 40 kg
  • Best proven commercial scale: UBTECH Walker S1/S2 (5,000+ units sold)
  • Best consumer entry point: 1X NEO at $499/month
  • Highest architectural ambition: XPENG IRON at 82 active / 200 total DOF

The humanoid robot market in mid-2026 is not one market — it’s four: research/education, consumer home, light industrial, and heavy manufacturing. Specs only translate to value when matched to deployment context. A Fourier GR-2 is overbuilt for a university lab; a Unitree G1 is underbuilt for an automotive line. The robots earning commercial traction — Digit, Apollo, Figure 03, Walker S1 — are winning not because their specs are highest, but because their specs are right for the environments they’re actually operating in.

For interactive side-by-side filtering across all platforms covered in this analysis, explore the full humanoid robot comparison tool or browse the complete 2026 robot directory. Deployment case studies and operator reviews are available at HumanoidApplications.com/deployments.

🌐