Amap launches ABot-Earth0.7: km-scale 3D cities in 10 min on a consumer GPU
智东西 · wechat · 2026-09-10
Amap (Gaode) unveiled its new 3D-native urban world model ABot-Earth0.7, alongside three AI services: Flight Street View 2.0, Navigation Live, and a trip risk-checker.
Key capabilities:
- Generates a km-scale 3D city scene in 10 minutes on a single consumer GPU from one satellite image or a text prompt — claimed 1000x efficiency gain over traditional pipelines
- 3D-native architecture trained on spatiotemporal data, outputting 3DGS scenes; one model spans planet-to-street scales, covering 196+ countries
Products:
- Flight Street View 2.0: freely explore a 3D destination before arriving, e.g. previewing theater seat sightlines before buying tickets
- Navigation Live: full-duplex voice, camera-based real-world perception, and continuous context across driving, walking, and scenic areas — positioned as a long-horizon spatial agent
- Trip checker: spatiotemporal simulation across 15 dimensions to flag schedule conflicts, congestion, and closures
Amap frames its spatial intelligence in three layers (3D representation, dynamic perception, spatiotemporal reasoning) and claims nearly 1 trillion daily BeiDou positioning calls. Mercedes-Benz signed on as its first global auto partner.
More from Models
- Nanbeige4.3-33B benchmarked: 59 tok/s single stream, ~1,060 tok/s across 64 concurrent runs — MaziyarPanahi · 2026-10-26
- Only 2 of 25 full ChatGPT rewrites flagged 100% AI by detectors — paulnovosad · 2026-10-26
- Qwen 4 Reportedly Spotted in the Arena, Hints at Alibaba's Next-Gen Model Testing — AIFlow_ML · 2026-10-26
- Nathan Lambert's first open models lecture: distillation, risks and the ecosystem explained — natolambert · 2026-10-26
- Nanbeige 4.3-33B flops real-world test: zero compilable output vs Qwen — julianharris · 2026-10-26
- Claude's safety classifier blocks SS7 defense research, pushing a security dev to an abliterated GLM — l4rz · 2026-10-26