gfs.ai

Tesla/Humanoid

Optimus

Tesla's bid to build humanoids by the million.

Announced$30,000Targeted H2 2026
  • Backed by Tesla's batteries, chips and factories
  • Targeting ~$30k, with talk of under $20k at volume
  • Gen 3 still unrevealed as of mid-2026
Non-gendered

Realism index

34/100

Our editorial estimate of how close this machine is to a lifelike human companion — across face, skin, movement and mind. How we score

No machine in this directory carries more expectation, or more asterisks, than Optimus. Tesla's pitch is intoxicating: a humanoid built on the same vision stack as its cars, made at automotive scale, eventually priced like a cheap sedan. If any of that lands, it resets the economics of the entire field.

What's verifiable in 2026 is more modest. The Gen 3 robot — the one with the new AI5 chip, the Grok voice, and the 22-degree-of-freedom hands — hadn't been officially revealed by mid-year. Low-volume production at Fremont kept sliding. And while Tesla spoke of a thousand robots being built internally, independent reporting found none doing genuinely productive factory work.

We include Optimus not because it's a companion — it has no face and no warmth — but because it's the category's gravity well. Every other maker prices against the day Tesla either delivers a sub-$20k humanoid or admits it can't. Until the production robot is real and working, treat the specs here as Gen 2 fact plus Gen 3 targets, and treat the timeline as aspiration.

Our verdict

The case for

  • Tesla's manufacturing scale and vertical integration
  • The most aggressive cost-down target in the field
  • A vast in-house factory testbed to learn on

The case against

  • Repeatedly delayed — the Gen 3 reveal kept slipping
  • No genuinely productive factory work as of early 2026
  • No external customers announced

What it can do

  • Factory pick-and-place and parts handling (supervised)
  • Bipedal walking with basic manipulation
  • Conversational voice (demonstrated)