---
title: SAM 3 on SIE | Superlinked
description: SIE, an open-source inference engine, serves SAM 3, an open detector from Meta that puts a box around every object you name in a photo.
canonical_url: https://superlinked.com/models/facebook-sam3
last_updated: 2026-11-07
---

# SAM 3

An open detector from Meta that puts a box around every object you name in a photo.

Model id: `facebook/sam3` · Endpoint: `extract` · [Hugging Face](https://huggingface.co/facebook/sam3)

Listed under [Detect](/models#detect) on [the model index](/models.md).

| Accuracy | Time per request | Serverless price |
|---|---|---|
| – (COCO, average precision · not measured) | 82.2 ms per photo | $0.25 per 1,000 images |

## On the task pages

- [Find objects more accurately than GPT‑6.1 Sol, 982x faster](/detect): 0.62 Objects boxed correctly; 82.2 ms Time per photo; $0.25 Cost per 1,000 photos

## Call it

On SIE Serverless:

```python
from urllib.request import urlopen
from sie_sdk import SIEClient

client = SIEClient(api_key="sk-sie-…")

image_bytes = urlopen("https://huggingface.co/datasets/superlinked/sie-task-evidence/resolve/0031c63f8da106f93dc6537d38f03139b7a0bb44/detect-api-input/2026-10-14/junction.jpg").read()
image = {"data": image_bytes, "format": "jpeg"}
result = client.extract(
    "facebook/sam3",
    {"images": [image]},
    labels=[
        "car",
        "bus",
        "truck",
    ],
)
for obj in result["objects"]:
    print(obj["label"], obj["score"], obj["bbox"])
```

## Next to the other detectors

| Model | COCO, average precision | Time per request | Serverless price | Size |
|---|---|---|---|---|
| [SAM 3](/models/facebook-sam3.md) `facebook/sam3` | – | 82.2 ms per photo | $0.25 per 1,000 images | 860M |
| [grounding dino base](/models/idea-research-grounding-dino-base.md) `IDEA-Research/grounding-dino-base` | 0.486 | – | $0.31 per 1,000 images | 233M |
| [owlv2 base patch16 ensemble](/models/google-owlv2-base-patch16-ensemble.md) `google/owlv2-base-patch16-ensemble` | 0.434 | – | $0.233 per 1,000 images | 155M |
| [owlv2 large patch14 ensemble](/models/google-owlv2-large-patch14-ensemble.md) `google/owlv2-large-patch14-ensemble` | 0.428 | – | Not on Serverless (run in your cloud) | 438M |
| [grounding dino tiny](/models/idea-research-grounding-dino-tiny.md) `IDEA-Research/grounding-dino-tiny` | 0.403 | – | Not on Serverless (run in your cloud) | 172M |

– = not measured / not timed / not listed.

[All 8 detectors](/models#detect)

## Facts

- Size: 860M parameters
- License: other
- Reads: Image
- Endpoint: extract

## Evaluation

- Accuracy: Run on SIE against the public dataset named in each figure
- Time: SIE timed on a GPU server with the client alongside, one request at a time
- Prices: Serverless list prices
- Sources: [Task page runs and their inputs](https://huggingface.co/datasets/superlinked/sie-task-evidence)