How we score a humanoid
Eight capability axes, broken into 40 sub-axes. Each sub-axis is anchored 0–5 and tied to a piece of evidence with a label.
A published score is the levelled 8-axis read, capped by the strength of the commercial evidence behind it. Sub-axes with no evidence count as 0 against a fixed 40 sub-axis denominator.
8 axes · 40 sub-axes
Every sub-axis has a 0–5 anchor descriptor: the wording an analyst has to see in the evidence before that score can be awarded. Specs and demos are recorded, but they are not scored as capability.
Steady gait on level surfaces with payload.
Anchor descriptors for this sub-axis are published with the first Cyborg Index release on 31 July 2026.
Every datapoint carries a label
The UI never visually equates a vendor claim with an independently validated observation. Eight labels, applied at the row level.
Six stages from claim to scale
Readiness is read as a position on this runway. An axis cannot score above the stage its commercial evidence supports.
- Stage 1Claim
- Stage 2Demo
- Stage 3Pilot
- Stage 4Named deployment
- Stage 5Measured operation
- Stage 6Scale
Commercial relationships have no influence over rankings, evidence labels, deployment validation or the methodology.
Methodology revisions are appended with each monthly Index release. If a score or a deployment is revised, the change is logged rather than replaced in place.
The Cyborg Index is published openly and is free to use and teach with attribution and a link back. Full terms on the licence page.