Humanoid Benchmark
Independent humanoid-robot index · Xodexa
XHI v2.0.0
← Back to the Humanoid leaderboard

🇨🇳 Galbot G1

Galbot · China · unveiled 2024
commercial task-autonomous electric data confidence: medium
69.7
XHI Score
Rank #18 of 156 humanoid

Score breakdown

AutonomyManipulationMobilityHardwareCommercialEcosystem
Autonomy & Intelligence · 25%92.0
Manipulation & Dexterity · 20%46.7
Mobility & Locomotion · 15%72.5
Hardware & Engineering · 15%42.4
Commercial Readiness & Deployment · 15%100.0
Ecosystem & Viability · 10%51.0

Each pillar is scored 0–100 from published specs, then weighted into the XHI. Compare it only with other humanoid robots — each category uses its own reference scales. See the full methodology →

Notable

Wheeled-base (omnidirectional) humanoid with a torso that lifts ~65 cm to reach 0-240 cm and ~10 h endurance — not a bipedal walker. Raised $300M+ in Dec 2025 (total ~$800M) at a $3B valuation.

Deployments & partners

  • FamilyMart (Beijing)
  • Galbot Store (autonomous retail, 30+ cities)
  • Manufacturing
  • Logistics
  • Healthcare
  • 2026 CCTV Spring Festival Gala

Specifications

Height
173.0 cm
Weight
85.0 kg
Payload
5.0 kg
Walking speed
Runtime
10.0 h
Battery
Total DoF
Hand DoF (per hand)
Dexterous hands
Yes
Actuation
electric
Form factor
bipedal
Onboard compute
AI model
AstraBrain
Indicative price
$98,000
Company valuation
$3.0B

Embed this score

XHI score: 69.7

Change history · 1 applied update

  • Spec update Update Galbot G1
    status: pilot → commercial; ai_model: Large-model general-purpose manipulation (embodied AI) → AstraBrain; dexterous_hands: False → True; form_factor: wheeled-base → bipedal; price_usd: 90000 → 98000; deployments: Galbot Store (autonomous retail, 30+ cities), Manufacturing, Logistics, Healthcare → FamilyMart (Beijing), Galbot Store (autonomous retail, 30+ cities), Manufacturing, Logistics, Healthcare, 2026 CCTV Spring Festival Gala
    2026-08-23T01:04:50+00:00 · by agent-bulk
Data confidence: medium. Some figures for this robot rest on single or conflicting sources, or on manufacturer/aggregator material rather than independent reporting. Treat second-order specs as approximate. Fields marked “—” were genuinely undisclosed and left blank rather than estimated.