Detectivision checks content for you. Send a piece of text, an image or a video (the file itself, or its address), the address of a recording, or start a live session, and get back which flags it raised and how sure the models are. In text and images every flag judges the whole text or image: answers have no positions, and capture and censor return what detect returns, at the same price.
It is a serverless API: there is nothing to install or scale. You call one HTTPS endpoint with an API key and pay only for completed checks.
curl https://api.detectivision.ai/api/v1/moderation \
-H "Api-Key: $DETECTIVISION_API_KEY" \
-H "Content-Type: application/json" \
-d '{"type": "image_moderation", "input": "https://example.com/photo.jpg"}'import os, requests
r = requests.post(
"https://api.detectivision.ai/api/v1/moderation",
headers={"Api-Key": os.environ["DETECTIVISION_API_KEY"]},
json={"type": "image_moderation", "input": "https://example.com/photo.jpg"},
)
print(r.json()["flagged"])const res = await fetch("https://api.detectivision.ai/api/v1/moderation", {
method: "POST",
headers: { "Api-Key": process.env.DETECTIVISION_API_KEY, "Content-Type": "application/json" },
body: JSON.stringify({ type: "image_moderation", input: "https://example.com/photo.jpg" }),
});
const { flagged, flags } = await res.json();What you can moderate#
| Type | Input | What comes back |
|---|---|---|
| Text | The text itself, up to 10,000 characters (English, Spanish, Portuguese, French, German, Italian, Turkish, Arabic, Russian, Hindi, Chinese and Japanese) | The flags raised, each with its confidence. |
| Image | The image as a file, its URL or base64 (JPEG, PNG, WebP, BMP, GIF, up to 10 MB) | The flags raised, each with its confidence. |
| Video | The video as a file (up to 20 MB) or its URL (up to 100 MB): MP4, MOV, WebM, MKV, AVI | The flags raised, each with its confidence, and when and where each finding is. |
| Audio | The address of a recording: a podcast episode, a voice message, a call | The flags raised, each with its confidence and the start and end of each finding in seconds. |
| Live stream | A stream link (RTSP, RTMP, HLS, HTTP, SRT) or frames from your camera | The flags raised on each frame, and the flagged moments, kept with their picture. |
Key concepts#
- Moderation type. What kind of content you send:
text_moderation,image_moderation,video_moderationoraudio_moderation; live streams are checked in live sessions. Each type has its own credits. - Flag. A category the checks look for. Your account's flags depend on its plan;
GET /api/v1/flagslists them. A result lists the flags that were raised. See Flags. - Mode. How much you want back:
detect,capture(the default) orcensor. For text and images all three return the flags and their confidence, at the same price. See Modes and censoring. - Credits. Each completed check costs credits from the type's balance; a check that could not finish costs nothing. See Credits and billing.
- Result. Every check returns the same shape:
flagged, theflagswith theirdetections, the flagschecked, andusage. See Responses.
Where to go next#
- Quickstart: an API key and your first check in five minutes.
- Authentication: API keys and their permissions.
- API Reference: every endpoint and field.
- MCP server: let your coding agent read these docs.
- Service status: the state of each moderation type and flag.