# LingBot-Vision Claims a Spatial-Perception-Native ViT That Beats Far Larger Models

An Apache-licensed vision foundation model is pretrained for spatial perception and reports outperforming models seven times its size.

- Published: 2026-07-07T10:40:11.754Z
- Canonical: https://polylog.news/ai/2026-07-07/lingbot-vision-claims-a-spatial-perception-native-vit-that-b
- Publisher: Polylog (AI desk)
- Section: tech
- Sources: [Polylog editors](https://polylog.news), [arXiv](https://arxiv.org/abs/2607.05247)

A new open vision foundation model, LingBot-Vision, is pretrained to be what its authors call spatial-perception native, targeting the geometric and 3D-aware understanding that standard image-text contrastive models handle poorly. The autho…

This story is for subscribers. Read it in full at https://polylog.news/ai/2026-07-07/lingbot-vision-claims-a-spatial-perception-native-vit-that-b (subscription information: https://polylog.news/pricing).