Amap (Alibaba) · 2026-06-08 · major
Amap Open-Sources ABot-Earth 0.5 — Alibaba's 3D Native World Model Generates Kilometer-Scale 3D Gaussian-Splatting City Scenes From a Single Satellite Image or Text Prompt in About 10 Minutes per km² on a Consumer GPU
Amap's CV Lab released ABot-Earth 0.5, a generative 3D world model that turns satellite imagery or text into 3DGS urban scenes in roughly 10 min/km². Code, tech report, and Studio beta are live; Amap already uses it to train the Amap Tutu robot.
Alibaba's Amap turns one satellite image into an interactive 3D city in roughly 10 minutes per square kilometer.
Key specs
| GitHub stars | 109 |
|---|---|
| Generation time | ~10 min / km² |
| Output format | 3D Gaussian Splatting |
| Hf papers upvotes | 193 |
| Claimed speed up vs traditional pipelines | ~1000× |
What is it?
ABot-Earth 0.5 is a generative 3D world model from Amap's CV Lab at Alibaba Group. It synthesizes seamless 3D urban scenes from geospatially-referenced satellite imagery — or a text prompt — and outputs them as 3D Gaussian Splatting scenes that can be loaded straight into Unity, Unreal, or the team's own web viewer.
How does it work?
The model is trained directly on 3D data so it has an intrinsic understanding of three-dimensional space rather than relying on per-scene NeRF optimization. A hierarchical level-of-detail structure streams large areas to a web mapping platform for interactive preview, and end-to-end generation runs on a single consumer-grade GPU at roughly 10 minutes per square kilometer.
Why does it matter?
Amap claims a 1,000× speed-up and about 1% of the cost compared with traditional scene-building pipelines. It is already shipping inside Amap, where it cut virtual-environment construction for the Amap Tutu autonomous quadruped from days to minutes — a concrete proof point that 3DGS world models are becoming usable training data for embodied AI rather than research demos.
Who is it for?
robotics, autonomous-driving, AR/VR and game teams that need photoreal city-scale simulation environments
Try it
git clone https://github.com/amap-cvlab/ABot-Earth-0.5