Nature WillChina2026

Natura-1

Key figures as stated by the developer

3 of 7 key figures stated

Controls ⁠A arms and hands
Inputs ⁠A camera language
Parameters ⁠A 50B

Not stated: Model type · Access · Base model · Licence

Full specifications

5 of 14 fields stated

The letter points to the source at the foot of the page. Fields without a figure are dashed: the developer does not state them — that is not the same as zero, and not the same as no.
Field	Value	Source and caveats
Release
Released	13 October 2026	[A] "Chinese startup Nature Will released its Natura-1 robot model on October 13" ⁠B
Input and output
Model type	not stated	Note Nature Will's own term: "Natura-1 is our first general foundation model for the physical world." ⁠A
Inputs	`camera` `language`	Developer wording: "It takes in language and vision and drives 20-DoF dexterous hands within a 100-millisecond control loop." Vision = camera, language = language. ⁠A
Controls	`arms and hands`	Developer wording: "It takes in language and vision and drives 20-DoF dexterous hands within a 100-millisecond control loop." Test platform: "On our dual-arm platform (ROKAE AR5 with Wuji hands)". ⁠A
Degrees of freedom controlled	not stated	Note The figure is per hand, not a count for the whole setup: "It takes in language and vision and drives 20-DoF dexterous hands within a 100-millisecond control loop." Test platform: "On our dual-arm platform (ROKAE AR5 with Wuji hands)". ⁠A
Size and speed
Parameters	50B	Developer wording: "Natura-1 is a 50-billion-parameter mixture-of-experts (MoE) model pretrained on more than one million hours of egocentric human experience." ⁠A
Control rate	not stated	Note The page gives a loop time, not a rate: "It takes in language and vision and drives 20-DoF dexterous hands within a 100-millisecond control loop." ⁠A
Where it runs and what it is built on
On robot or cloud	not stated
Base model	not stated
Openness
Access	not stated
Licence terms	not stated
Licence	not stated
The developer’s own claims
Learns new tasks	without examples	Developer claim: "It is a single model that follows plain language instructions zero-shot and carries them out with dexterous hands." The adaptation to a robot hand is a separate step: "we conduct efficient post-training on a limited amount of robot data to adapt pretrained world knowledge to dexterous manipulation". Six long-horizon tasks "use additional tuning on top of Natura-1; all other results share the same weights". ⁠A
Works across robots	not stated	Note The page's tests use one platform with dexterous hands: "On our dual-arm platform (ROKAE AR5 with Wuji hands)". ⁠A Yes only if the developer itself says the model runs on different robot models, for example robots from two manufacturers or two of one manufacturer’s models. Several units of the same robot do not count.

How to read the figures

50HzA: The letter in the grey square after a figure points to the source the figure came from. Pages with several sources carry several letters.
50HzB: A dashed letter marks a secondary source – a datasheet, manual or developer documentation from the manufacturer, or a press report used only where the manufacturer gives no figure.

What the data labels mean

Notes

Nature Will describes Natura-1 as its first general foundation model for the physical world.
The Nature Will website footer carries a Beijing ICP registration number.
humanoid.guide reports that Nature Will released Natura-1 on 13 October 2026.

Versions and notes in the key figures (4)

Model type

Note Nature Will's own term: "Natura-1 is our first general foundation model for the physical world."

Controls

Developer wording: "It takes in language and vision and drives 20-DoF dexterous hands within a 100-millisecond control loop." Test platform: "On our dual-arm platform (ROKAE AR5 with Wuji hands)".

Inputs

Developer wording: "It takes in language and vision and drives 20-DoF dexterous hands within a 100-millisecond control loop." Vision = camera, language = language.

Parameters

Developer wording: "Natura-1 is a 50-billion-parameter mixture-of-experts (MoE) model pretrained on more than one million hours of egocentric human experience."

Sources

A Primary source. www.naturewill.ai/blog/natura-1 · Retrieved 2026-10-19
B Secondary: humanoid.guide/nature-will-unveils-natura-1-robot-model-for-dexterous-hands/ · Retrieved 2026-10-19

One line per source, not per figure: a source has one retrieval date. The letter beside a figure points to the line it came from.