# Open-Vocabulary, Promptable Vision Foundation Models

Vision foundation models shift to text-promptable, open-vocabulary detection, segmentation, and real-time tracking of arbitrary concepts, generalizing perception beyond fixed label sets across images and video and pushing open perception models toward production use.

- Conviction: 96 / 100 (strengthening)
- 7-day move: +17
- Horizon: Short term (next 30 days)
- Tracking since: 2026-06-16T00:00:00.000Z
- Last updated: 2026-07-07T14:00:02.329Z
- Canonical: https://polylog.news/ai/trends/open-vocabulary-vision-foundation-models
- Publisher: Polylog
- Affected regions: Global

## Recent score history

- 2026-07-06: 93
- 2026-07-07: 96

## Recent evidence

- [confirms] LingBot-Vision Claims a Spatial-Perception-Native ViT That Beats Far Larger Models (2026-07-07): The Apache-licensed LingBot-Vision, a spatial-perception-native ViT, reports outperforming vision models seven times its size, per the tech report. Another open, production-oriented vision foundation model advances the shift of perception onto general-purpose foundation models.
- [confirms] Segment Anything Underpins a Fashion App, Showing Open-Vocabulary Vision in Production (2026-07-06): Fashion app Alta Daily used Meta's Segment Anything to process more than 20 million clothing images, a production example of promptable, open-vocabulary segmentation moving from research to product.

14 more evidence entries, the full score history, the conviction-driver timeline, and affected assets are for subscribers: https://polylog.news/pricing