AgibotChina2026
GE-Sim 2.0
An AI model: a vision-language-action model (VLA) takes camera images and an instruction and turns them into movements for a robot. This page describes the software, not a robot.
Robotsnot stated
Key figures as stated by the developer
4 of 6 stated · 2 gaps- Model type not stated
- Controls not stated
-
Inputs A
camera - Access A open download
- Parameters B 2,000,000,000
- Base model C Cosmos-Predict2-2B-Video2World
How to read the figures
- 14kgA30.9lbA
- The letter in the grey square after a figure points to the source the figure came from. Robots with several sources carry several letters.
- 74.5lb
- Imperial figures are our own conversion unless the manufacturer states an imperial figure itself. The source letter always belongs to the metric figure — a conversion has no source of its own.
Versions and notes in the key figures (6)
Model type
Note The developer calls it something else and does not label it vision-language-action or world action model: "a closed-loop video world simulator for robotic manipulation".
Controls
Note The model generates video and robot state for simulation; it does not itself command a robot: "Given multi-view history frames and an action trajectory, the model generates multi-view future rollouts that follow the specified robot behavior".
Inputs
"Given multi-view history frames and an action trajectory, the model generates multi-view future rollouts that follow the specified robot behavior".
Access
On Hugging Face: "The released GE-Sim 2.0 world simulator checkpoint."
Parameters
The size given in the abstract: "at only 2B parameters".
Base model
The paper instantiates the simulator with the "Cosmos-Predict2-2B-Video2World DiT".
Full specifications
4 of 12 fields stated| Field | Value | Source and caveats |
|---|---|---|
| Input and output | ||
| Model type | not stated | Note The developer calls it something else and does not label it vision-language-action or world action model: "a closed-loop video world simulator for robotic manipulation". B |
| Inputs | camera | A "Given multi-view history frames and an action trajectory, the model generates multi-view future rollouts that follow the specified robot behavior". |
| Controls | not stated | Note The model generates video and robot state for simulation; it does not itself command a robot: "Given multi-view history frames and an action trajectory, the model generates multi-view future rollouts that follow the specified robot behavior". A |
| Degrees of freedom controlled | not stated | |
| Size and speed | ||
| Parameters | 2,000,000,000 | B The size given in the abstract: "at only 2B parameters". |
| Control rate | not stated | |
| Where it runs and what it is built on | ||
| Runs where | not stated | Note The paper reports "a 25-frame rollout in 2.3 seconds on a single H100". B |
| Base model | Cosmos-Predict2-2B-Video2World | C The paper instantiates the simulator with the "Cosmos-Predict2-2B-Video2World DiT". |
| Openness | ||
| Access | open download | A On Hugging Face: "The released GE-Sim 2.0 world simulator checkpoint." |
| License | not stated | Note The licence section does not name the weights; it says "Other data and codes within this repo are under" CC BY-NC-SA 4.0; code adapted from Diffusers and Cosmos is under Apache License 2.0. A |
| The developer’s own claims | ||
| Learns new tasks | not stated | |
| Works across robots | not stated | Note The project page says it "enables precise action-conditioned video generation across diverse robotic embodiments"; on Hugging Face: "It is post-trained on Genie-01(G01) + OmniPicker(OP) data". 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. |
Notes
- AgiBot calls GE-Sim 2.0 (Genie Envisioner World Simulator 2.0) a closed-loop video world simulator for robotic manipulation.
- AgiBot announced Genie Envisioner 2.0 as the step from World Action Models to interactive world simulators.
- The technical report was submitted to arXiv on 26 May 2026.
Sources
-
A
Manufacturer product page. huggingface.
co/ · Retrievedagibot-world/ Genie-Envisioner-Sim-v2. 0 -
B
arxiv.
org/ · Retrievedabs/ 2605. 27491 -
C
arxiv.
org/ · Retrievedhtml/ 2605. 27491
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.