# Meta's Segment Anything 3 Makes Open-Vocabulary Segmentation a Text Prompt

The model detects, segments, and tracks every instance of a concept from a phrase like "yellow school bus," roughly doubling prior systems on Meta's open-vocabulary benchmark.

- Published: 2026-06-22T10:48:41.833Z
- Canonical: https://polylog.news/ai/2026-06-22/meta-s-segment-anything-3-makes-open-vocabulary-segmentation
- Publisher: Polylog (AI desk)
- Section: tech
- Sources: [Meta AI](https://ai.meta.com/blog/segment-anything-model-3/), [Meta AI (Alta Daily case study)](https://ai.meta.com/blog/alta-daily-fashion-app-segment-anything/)

Meta's Segment Anything Model 3 (SAM 3) shifts the family from pixel-level prompting to concept-level prompting. Where SAM and SAM 2 required a manual click or box per instance, SAM 3 takes an open-vocabulary text description and segments a…

This story is for subscribers. Read it in full at https://polylog.news/ai/2026-06-22/meta-s-segment-anything-3-makes-open-vocabulary-segmentation (subscription information: https://polylog.news/pricing).