Benchmarks rarely look like this. LingBot-Map, the new feed-forward 3D reconstruction foundation model from Ant Group’s Robbyant team, scores 98.98 F1 on ETH3D — more than 21 points above second place. It hit #1 on GitHub trending with +827 stars in a single day, and both code and weights are open.
What it actually does
Point a plain RGB camera at the world, walk around, and get a live 3D map. No LiDAR, no depth sensor, no offline processing. It runs at roughly 20 FPS and streams through 10,000+ frame videos without accuracy falling apart — the classic failure mode of streaming reconstruction. The trick is a pure auto-regressive design with Geometric Context Attention that keeps geometry consistent across frames.
This is the perception layer for robots, autonomous driving, and AR glasses. It’s also the newest member of the LingBot family (Depth, VLA, World) — Ant is quietly assembling a full embodied AI stack, one open-source piece at a time.
How to run it
No hosted API. Grab the open code, pull the weights from Hugging Face, and deploy locally — a GPU plus any RGB video stream is enough to build real-time spatial understanding into your own robot or app.
You Might Also Like
- Heretic Just hit Github Trending and the ai World has Opinions
- Pentagi Just hit 1 on Github Trending and Yeah its Worth the Hype
- Ant Group Alipay ai pay aq Health Just hit 100 Million Users Each and Thats a big Deal
- Pageindex Just hit Github Trending and it Might Make you Rethink rag Entirely
- Mirofish Just hit Github Trending and its Unlike any ai Tool ive Seen
Top AI Product 
Leave a comment