Facial verification, without the infrastructure.

Compare two faces with one API call. We handle hosting, scaling and model updates.

Try an example

Use with

Codex

Claude Code

Cursor

50 calls free each month. No card needed.

Send a photo.
Get an embedding back.

curl https://api.deepface.dev/represent \
  -H "x-api-key: $DEEPFACE_API_KEY" \
  -F "img=@face.jpg"

Three ways to work with faces.

Explore the inputs, the vectors and the results.

Compare two photos.

Send two photos to check whether they show the same person.

Example

01Start with two photos

Studio photo of a fictional adult — Studio photo

Outdoor photo of a fictional adult — Outdoor photo

POST /verifyPhotos in. Match decision out.

02ResultPrecomputed

Face match

These photos pass the match threshold.

Cosine distance: 0.0504
Match threshold: ≤ 0.40

Lower distance means more similar faces.

View JSON

{
  "confidence": 100,
  "detector_backend": "opencv",
  "distance": 0.050446,
  "facial_areas": {
    "img1": {
      "h": 299,
      "left_eye": [
        365,
        242
      ],
      "right_eye": [
        262,
        244
      ],
      "w": 299,
      "x": 167,
      "y": 126
    },
    "img2": {
      "h": 294,
      "left_eye": [
        371,
        235
      ],
      "right_eye": [
        268,
        243
      ],
      "w": 294,
      "x": 178,
      "y": 123
    }
  },
  "model": "Facenet",
  "similarity_metric": "cosine",
  "threshold": 0.4,
  "time": 0.48,
  "verified": true
}

Start with your free calls.

50calls / month

Add a card for 50 more.

Pay as you go when you need more calls.

How pricing works Talk to an engineer

Live onboarding

Let’s run your first request together.

Bring your workflow and we’ll help you make your first API request.

Join an onboarding sessionAPI walkthrough + Q&A