Delta IntelligenceChina2026
Delta-0
Full specifications
4 of 14 fields stated| Field | Value | Source and caveats |
|---|---|---|
| Release | ||
| Released | 28 September 2026 | [A] "2026.09.28 Delta Intelligence Unveils Delta-0, a Native Humanoid Robot Foundation Model" B |
| Input and output | ||
| Model type | not stated | Note Delta Intelligence's own term: "Δ₀, Delta Intelligence’s humanoid foundation model (HFM) for whole-body loco-manipulation". Its brain: "Δ₀ pairs a brain—a latent world-action model—with a learned whole-body controller." A |
| Inputs | camera language robot state | Developer wording: Delta-0 works "using cameras, language, and the robot’s own state to coordinate movement and manipulation". Cameras = camera. A |
| Controls | whole body | Developer wording: "Δ₀ brings the hands, arms, torso, and legs into a shared action space". A |
| Degrees of freedom controlled | not stated | Note The blog gives the figure without naming a robot model: "Unlocks a humanoid’s full dexterity across 69 DoFs". A |
| Size and speed | ||
| Parameters | not stated | |
| Control rate | not stated | |
| 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 | needs retraining | Developer claim: Delta-0 learns from human data at scale, "building a foundation for coordinated behavior and transfer to downstream tasks". Post-training: "Real-world reinforcement learning extends policy training beyond demonstrations by learning from autonomous attempts and targeted human guidance." A |
| Works across robots | not stated | Note The blog names no robot models; it describes training data from several embodiments: "We propose a shared action representation for Δ₀ that enables the brain to learn from different data sources and embodiments." 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.
Notes
- Delta Intelligence describes Delta-0 as a humanoid foundation model for whole-body loco-manipulation.
- Delta Intelligence unveiled Delta-0 on 28 September 2026.
- Delta Intelligence lists an address in Haidian District, Beijing.
Versions and notes in the key figures (3)
Model type
Note Delta Intelligence's own term: "Δ₀, Delta Intelligence’s humanoid foundation model (HFM) for whole-body loco-manipulation". Its brain: "Δ₀ pairs a brain—a latent world-action model—with a learned whole-body controller."
Controls
Developer wording: "Δ₀ brings the hands, arms, torso, and legs into a shared action space".
Inputs
Developer wording: Delta-0 works "using cameras, language, and the robot’s own state to coordinate movement and manipulation". Cameras = camera.
Sources
-
A
Primary source. deltai.
com/ · Retrieveden/ blog/ delta-0 -
B
deltai.
com/ · Retrieveden/ news
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.