# Berkeley Lab fine-tunes Meta's SAM 3 and DINOv3 for Energy Department beamline work under the Genesis Mission

Meta says segmentation that took about a month of expert effort per time step now takes roughly 15 minutes per reconstructed 3D volume. The lab's own pages do not give that figure.

- Published: 2026-11-22T06:11:42.557Z
- Canonical: https://polylog.news/ai/2026-11-22/berkeley-lab-fine-tunes-meta-s-sam-3-and-dinov3-for-energy-d
- 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/), [Lawrence Berkeley National Laboratory](https://www.lbl.gov/genesis-mission-projects/), [Runtime Wire](https://runtimewire.com/article/berkeley-lab-meta-models-doe-genesis-mission)

Meta described how Lawrence Berkeley National Laboratory uses its open-weight SAM 3 and DINOv3 models in SYNAPS-I, one of the first projects under the Department of Energy's (DOE) Genesis Mission. The Advanced Light Source (ALS) at Berkeley…

This story is for subscribers. Read it in full at https://polylog.news/ai/2026-11-22/berkeley-lab-fine-tunes-meta-s-sam-3-and-dinov3-for-energy-d (subscription information: https://polylog.news/pricing).