How to read a humanoid ranking
The Cyborg Index reads the public evidence record of humanoid robots, not the marketing. This lesson walks through one real robot so you can see what the numbers mean — and what they do not mean.
One sentence to remember
A high rank means a robot has the strongest publicly evidenced capability record right now. It does not mean it is the "best" robot, the most advanced in secret, or the one that will win in five years.
Meet Tesla Optimus
Tesla Optimus is ranked #10 in the August 2026 edition. It is famous, but fame is not evidence. Here is how its scores break down.
Average across 40 capability checks. This is what ranks the robot.
How much of that score is backed by public evidence.
How many checks have any public evidence at all.
Tesla Optimus's Capability score is 0.72, but its Proof score is only 0.16. That gap means most of its recorded capability is supported by weaker sources — claims and demos — rather than independent or operational evidence.
The whole field is still early
The current leader, UBTech Walker S, has a Capability score of 1.48 out of 5. That is the top of the table — and it is still below 1.5. The ranking is not saying humanoids are almost ready. It is saying the public record is thin across the whole field.
Evidence is capped, not ignored
Not all sources are equal. A vendor claim or a staged demo supports less than an independent report, a customer trial, or a measured result. Weaker sources are capped, not ignored.
No public evidence yet. Counts as 0 in the average.
The maker says the robot can do it. We have not seen it.
Shown in a video or staged setting. Useful signal, but capped.
Independently observed, measured, or replayable.
Commercial proof moves in stages
An axis cannot score above the stage its commercial evidence supports. Here is the ladder, from weakest to strongest.
Claim
The vendor says it can do something.
Demo
It can be shown under chosen conditions.
Pilot
It is being tested in a real setting.
Named deployment
A named customer is using it for real work.
Measured operation
Its performance is tracked and reported.
Scale
Many sites, many shifts, many tasks.
What to remember
- 01Rank reflects the public evidence record, not fame, funding, or future promise.
- 02Scores are out of 5. A top score around 1.5 means the whole field is still early.
- 03Missing evidence counts as 0, not a verdict that the robot cannot do the thing.
- 04A robot can rank high while most of its score rests on thin proof. That is visible as a proof gap.
Key terms
- Capability score
- The Index score. An average of 40 capability checks, each scored 0–5. It is the ranking authority.
- Proof score
- How much of the Capability score is backed by public evidence. It is a reference read, never the rank.
- Coverage
- How many of the 40 checks have any public evidence at all.
- Proof gap
- The distance between Capability score and Proof score. A large gap means the public record is thinner than the score implies.
- Evidence ceiling
- A weaker source can only lift a score so far. A vendor demo cannot push a check as high as an independent measured result.
Try it on the live ranking
Open the current Cyborg Index, pick any robot, and look for the same four numbers: Capability score, Proof score, Coverage, and Proof gap. The story is always in the gap.