# facebookresearch/sam3

The repository provides code for running inference and finetuning with the Meta Segment Anything Model 3 (SAM 3), links for downloading the trained model checkpoints, and example notebooks that show how to use the model.

Repository: https://github.com/facebookresearch/sam3
Canonical: https://ross.abutalabs.com/products/sam3
Homepage: https://ai.meta.com/sam3/
Language: Python
License: NOASSERTION
License Family: other
Last push: 2026-08-26T11:37:20+00:00

## Health v2 (maintenance only)
Score: 63/100 (v2, computed 2026-09-02T17:46:02.011165+00:00)
- activity 99, release rhythm 35, longevity 29
- inputs: {"age_days": 412, "days_push": 7, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases, no_license
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 11487, forks 1739 (observed 2026-08-28T04:10:48.304852+00:00)

## What it is
Official code for Meta's Segment Anything Model 3 (SAM 3), a unified foundation model for promptable segmentation in images and videos. It supports inference and fine-tuning, model checkpoint downloads, and example notebooks, adding open-vocabulary concept-based segmentation of all object instances via text or visual prompts.

## Use cases
- segment all instances of an object described by a text phrase in an image
- segment and track objects across video frames
- segment objects using point, box, or mask prompts
- fine-tune SAM 3 on custom segmentation data
- download and run SAM 3 checkpoints for inference
- build open-vocabulary detection and segmentation pipelines

## When to choose
- you need state-of-the-art promptable image or video segmentation
- you want to segment every instance of an open-vocabulary concept from a text prompt
- you need a foundation segmentation model to fine-tune on your own data
- you want official checkpoints and example notebooks from Meta

## When to avoid
- you need a permissively licensed model for commercial redistribution without restrictions (SAM license applies)
- you need lightweight real-time segmentation on CPU-only devices
- you only need simple classic image segmentation without deep learning

## Facets
- artifact type: library
- maturity: active
- function: machine-learning, computer-vision, image-processing, video-processing, sdk
- domain: computer-vision, image-processing, machine-learning, artificial-intelligence
- platform: python, cross-platform
- tags: segment-anything, promptable-segmentation, open-vocabulary, object-tracking, foundation-model, inference, fine-tuning, meta-ai, video, gpu

## Member repositories
- facebookresearch/sam3 (main) score 63

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:10:48.304852+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-29T17:15:33.399384+00:00, confidence not recorded.
  - readme: https://github.com/facebookresearch/sam3 (fetched 2026-08-28T04:10:48.304852+00:00, sha abea3544c0a1)
  - homepage: https://ai.meta.com/sam3/ (fetched 2026-08-29T08:14:08.612794+00:00, sha 3f1f00296cae)
  - registry_pypi: https://pypi.org/pypi/sam3/json (fetched 2026-08-29T08:14:08.648378+00:00, sha a2bfce2f3610)
- Data as of 2026-08-30T08:39:29.467469+00:00.