# Meta's Open Vision Models Cut a Month of Beamline Annotation to Fifteen Minutes at Berkeley Lab

SAM 3 and DINOv3 run on 300 Nvidia A100 accelerators inside a Department of Energy science project, and the same pair runs on battery-powered hardware in a $41.5 million assistive robotics program.

- Published: 2026-08-10T06:21:58.349Z
- Canonical: https://polylog.news/ai/2026-08-10/meta-s-open-vision-models-cut-a-month-of-beamline-annotation
- Publisher: Polylog (AI desk)
- Section: tech
- Sources: [Meta AI](https://ai.meta.com/blog/genesis-mission-lawrence-berkeley-national-laboratory-segment-anything-dino/), [Meta AI](https://ai.meta.com/blog/assistive-robotics-university-of-pittsburgh-sam-dino/)

Two deployments published by Meta show where its open vision models are actually being used. At Lawrence Berkeley National Laboratory, the segmentation model SAM 3 and the self-supervised representation model DINOv3 run on 300 Nvidia A100 a…

This story is for subscribers. Read it in full at https://polylog.news/ai/2026-08-10/meta-s-open-vision-models-cut-a-month-of-beamline-annotation (subscription information: https://polylog.news/pricing).