The Cyborg Index · July 2026 edition

Chinese Humanoid RobotsJuly 2026

Humanoids from China-headquartered makers, read against exactly the same framework as everyone else — no separate scale, no regional allowance. Ranks shown are positions in the full cohort, so a robot's place here is directly comparable to the global table.

UBTech Walker S (UBTech) leads the China-headquartered humanoids in the July 2026 Cyborg Index, ahead of Agibot A2 (Agibot (Zhiyuan Robotics)) and Kepler Forerunner K2 (Kepler Robotics). 6 of the 12 humanoids in this edition are China-headquartered humanoids, scored on the same 40 capability measures as the rest of the cohort.

1NEW
Proof score0.13
Proof gap
+1.57
Capability score1.70
Coverage
29/40
Evidence strength8%
2NEW
Agibot A2
Agibot (Zhiyuan Robotics)
Proof score0.19
Proof gap
+1.36
Capability score1.55
Coverage
26/40
Evidence strength12%
3NEW
Kepler Forerunner K2
Kepler Robotics
Proof score0.06
Proof gap
+1.26
Capability score1.32
Coverage
26/40
Evidence strength5%
9NEW
Unitree H2
Unitree
Proof score0.05
Proof gap
+0.72
Capability score0.78
Coverage
17/40
Evidence strength6%
11NEW
Fourier GR-2
Fourier Intelligence
Proof score0.06
Proof gap
+0.51
Capability score0.57
Coverage
14/40
Evidence strength10%
12NEW
Proof score0.01
Proof gap
+0.46
Capability score0.47
Coverage
11/40
Evidence strength2%

Ranks are positions in the full 12-humanoid cohort, so they stay comparable with the complete Index and with side-by-side comparisons.

Cyborg IndexJuly 2026 editionCapability scoremethodology

Questions

Which Chinese humanoid robot ranks highest right now?
In the July 2026 edition, UBTech Walker S from UBTech holds the highest Capability score in this list. That is a reading of the public evidence record at sign-off, not a claim about undisclosed internal capability.
How is this list ordered?
By Capability score: each humanoid is read across eight capability axes and 40 sub-axes, each scored 0–5 from public evidence, then levelled so the same number means the same thing on every robot. Sub-axes with no public evidence score zero over a fixed denominator, so silence is never rewarded.
How often does this list change?
Monthly. Each edition is dated and kept as published; this page re-states itself from the newest signed-off edition rather than being rewritten by hand.
Are Chinese humanoids scored differently?
No. The framework, the evidence ceilings and the levelling are identical. The only difference on this page is the filter: it shows makers headquartered in China, and the rank shown is the robot's position in the full global cohort.

How the scores are built: methodology.