AI Detector API

Send text from your app or agent for AI detection. When there's a verdict you get a 0–100 score and a label; English and Portuguese can also get sentence scores. Pay per word, no subscription.

5,000 trial credits with your first key (verified, non-disposable email; one trial per address). Then $0.04 per 1,000 scored words on the Starter pack, less on larger packs.

Request
# essay.json holds the full 207-word sample: {"text": "Remote work has fundamentally transformed..."}
curl https://aicheckerhumanizer.com/api/v1/detect \
  -H "Authorization: Bearer $AICH_API_KEY" \
  -H "Content-Type: application/json" \
  -d @essay.json
Response (excerpt)
{
  "score": 100,
  "verdict": { "basis": "dominant_language", "language": "en", "words": 207, "label": "likely_ai" },
  "scoredWords": 207,
  "sentences": [
    { "start": 0, "end": 95, "text": "Remote work has fundamentally transformed the way modern organizations operate and communicate.", "score": 100, "color": "ai" },
    { "start": 96, "end": 238, "text": "In today's rapidly evolving business landscape, companies must adapt to new paradigms of collaboration that transcend geographical boundaries.", "score": 100, "color": "ai" }
  ],
  "creditsCharged": 207,
  "creditsRemaining": 249793
}

What the API does

POST /detect takes up to 10,000 words and answers in the same request. It's the same detector that runs on our site.

When there's a verdict, the response includes a 0–100 score and a label (likely AI, mixed or likely human), in verdict.label for one dominant language or verdict.parts when several languages each get their own. Sentences come back with their positions; in English and Portuguese they also have scores and colors, while Spanish and French get a whole-text verdict only. We add up the words in paragraphs detected as a supported language; scoring needs at least the lowest minimum among the detected languages (150 words if a paragraph is detected as English, 200 otherwise); below that, sentences is empty; once scoring runs, sentences can come back even when no language has enough scored words for a verdict.

You're charged only when there's a verdict, for the words that were scored. A verdict needs at least 150 words in English, or 200 in Spanish, Portuguese and French; shorter text costs nothing.

The same key and credit balance work for both APIs: detection answers in one request, and rewrites use a task ID. See the AI humanizer API, or try the AI detector in the browser before you write any code.

Quick start

  1. Sign in and create a key on your account page. It's shown once, so store it somewhere safe.
  2. Send it as a Bearer token with each request.
  3. Read the response fields in the API documentation.
Request
# essay.json holds the full 207-word sample: {"text": "Remote work has fundamentally transformed..."}
curl https://aicheckerhumanizer.com/api/v1/detect \
  -H "Authorization: Bearer $AICH_API_KEY" \
  -H "Content-Type: application/json" \
  -d @essay.json

Use it from an AI agent

The API is also an MCP server, so Claude, Cursor and other MCP clients can call it as tools: detect_ai_text, humanize_text, get_humanize_result and get_balance. Credits and limits are the same as the REST API.

Claude Code
claude mcp add --transport http aicheckerhumanizer \
  https://aicheckerhumanizer.com/api/mcp \
  --header "Authorization: Bearer ach_live_..."
Cursor and other clients that take a URL
{
  "mcpServers": {
    "aicheckerhumanizer": {
      "url": "https://aicheckerhumanizer.com/api/mcp",
      "headers": { "Authorization": "Bearer ach_live_..." }
    }
  }
}

Pricing

Credits are prepaid and last 12 months. Detection uses 1 credit per scored word when there's a verdict. For each chunk that's rewritten (a paragraph, or part of a long one), humanizing uses 5 credits per word of the original; chunks shorter than 10 words are billed as 10. For more than $200 at a time, contact us.

API credit packs
PackPriceCreditsHumanizingDetection
Starter$10250,00050,000 words, $0.20 per 1,000250,000 words, $0.04 per 1,000
Growth$501,500,000 (+20%)300,000 words, $0.167 per 1,0001,500,000 words, $0.033 per 1,000
Scale$2006,500,000 (+30%)1,300,000 words, $0.154 per 1,0006,500,000 words, $0.031 per 1,000
Humanizing word counts assume every rewritten chunk has at least 10 words.

API credit packs: email us within 7 days of payment to ask for a full refund. The credits used and reserved from that pack must add up to 5% or less. One API refund per email address. See the full refund rules.

Limits and languages

  • Detection: up to 10,000 words per request, in English, Spanish, Portuguese and French.
  • Humanizing: up to 3,000 words and 200 chunks per request, in English, Spanish, Portuguese and French.
  • Per hour: 2,000 detections, 600 humanize requests and 20,000 result or balance checks. The response tells you how long to wait when you hit one.
  • At once: before your first credit purchase, 2 detections and 1 rewrite in progress; after it, 10 of each. Queued rewrites count.
  • Up to 5 active keys per account. Revoking a key stops it right away.

Data handling

We don't save the text you send for detection in our database. Passage fingerprints and scores are cached for 24 hours, as on our website. If you send an Idempotency-Key, we also keep the response scores and positions for 24 hours so a retry gets the same result. Humanizing results can be fetched for about 70 minutes after you submit; the text is stored encrypted and deleted at the next cleanup after that, at least 10 minutes after the task finishes. Your usage log keeps dates, word counts and credits for 90 days, never the text. The model providers we send text to have their own retention periods, listed in our privacy policy.

Frequently asked questions

Do I need a subscription?

No. Buy a credit pack once and use it within 12 months. Your first API key gets 5,000 trial credits for 30 days if your email is verified and isn't a disposable address; each address can claim the trial once. Website plans and API credits are separate, so a website subscription doesn't include API use.

Can I use the same credits for humanizing?

Yes. One balance covers both. Detection uses 1 credit per scored word when there's a verdict; humanizing uses 5 credits per original word of each rewritten chunk, and chunks shorter than 10 words are billed as 10.

What happens if I send the same request twice?

Without an Idempotency-Key, each submission is a new request. With one, retry within 24 hours using the same key, text and options: a completed detection replays without a second charge, one still running returns 409, and a failed one can run again. For an accepted rewrite the key returns the same task, whose text expires about 70 minutes after you submit.

Can I build it into a product I sell?

Yes. You can add it as a feature of your own app or use it on your clients' text. Don't share your key outside your organization or resell credits.

Do you keep the text I send?

We don't save detection text in our database. Humanizing results can be fetched for about 70 minutes after you submit; the encrypted text is deleted at the next cleanup after that, at least 10 minutes after the task finishes. The model providers we send text to keep it for their own retention periods, listed in our privacy policy. We don't use your text to train models.