Last updated
AI Detection API.
An AI detector API that scores every sentence for AI and returns the ones that read as AI, in under 2 seconds.
- $0.02 per 1,000 words
- A score for every sentence
- Under 2 seconds

From the team behind Phrasly.ai, as featured in
What is an AI detection API?
An AI detection API is a web service that checks whether text was written by AI. Your app sends the text in an HTTP request and gets back a score, usually an overall AI percentage and, with some APIs, a score for each sentence, so you can flag AI writing inside your own product.
Phrasly's AI detection API tells you whether text was written by a person or generated by a model such as ChatGPT, Claude or Gemini. It's the engine that powers phrasly.ai, as one REST endpoint. To check text by hand, use the Phrasly AI detector.
How the Phrasly AI detector API works.
Send one POST request with at least 50 words of text. The API splits it into sentences, scores each one with Phrasly's in-house detection model, and combines the scores into a document-level AI percentage.
- 1
Create a key
Subscribe on the pricing page and generate an API key in your dashboard.
- 2
Send text
POST at least 50 words to
/api/v1/detectwith your key in thex-api-keyheader. - 3
Get the result
A label, AI and human percentages, the flagged sentences and a score for every sentence, as JSON.
curl -X POST https://business.phrasly.ai/api/v1/detect \
-H "x-api-key: YOUR_API_KEY" \
-H "Content-Type: application/json" \
-d '{"text": "Text you want to check for AI..."}'import requests
response = requests.post(
"https://business.phrasly.ai/api/v1/detect",
headers={
"Content-Type": "application/json",
"x-api-key": "YOUR_API_KEY",
},
json={"text": "Text you want to check for AI..."},
)
data = response.json()["data"]
print(f"Label: {data['Label']}, AI: {data['AI']}%")const response = await fetch("https://business.phrasly.ai/api/v1/detect", {
method: "POST",
headers: {
"Content-Type": "application/json",
"x-api-key": "YOUR_API_KEY",
},
body: JSON.stringify({ text: "Text you want to check for AI..." }),
});
const result = await response.json();
console.log(result.data.Label); // "AI", "Might be AI", or "Human"
console.log(result.data.AI);Response
200application/json
{
"data": {
"Label": "AI",
"AI": 85.9,
"Human": 14.1,
"Chunks": ["..."],
"SentenceScores": [
{ "text": "...", "score": 0.57 },
{ "text": "...", "score": 0.22 }
],
"original": "..."
},
"message": "AI detection results are ready."
}| Field | Type | What it returns |
|---|---|---|
Label | string | Classification: "AI", "Might be AI" or "Human" |
AI | number | AI confidence percentage (0–100) |
Human | number | Human confidence percentage (0–100) |
Chunks | string[] | Sentences flagged as AI-generated |
SentenceScores | {text, score}[] | A 0–1 AI score for every sentence in the input |
original | string | The original input |
Errors and limits are in the detect endpoint reference.
How fast is the AI detector API?
Responses come back in under 2 seconds, fast enough for live products.
For volume, call it from your own scripts or pipeline, or scan a CSV of up to 100 documents at once in the dashboard.

A score for every sentence
Every sentence gets a 0–1 AI score in SentenceScores, so you can set your own thresholds or build heat-maps.
Mixed-authorship documents
Pinpoint the AI parts of a document that's only partly AI, which is common in collaborative writing and editing.
Cached re-scans
Repeated sentences are served from a cache, so re-scanning a lightly edited draft is fast.
Built for volume
Call it from your own scripts or pipeline, or scan a CSV of up to 100 documents in the dashboard.
Long documents
Split anything over 15,000 characters at paragraph breaks, send each section, and combine the sentence scores.
99.9% uptime SLA
Runs on Cloudflare's global network, and planned maintenance is announced ahead of time.
AI detection API pricing: $0.02 per 1,000 words.
The $100/month Business plan includes $100 in API credits every month, which covers up to 5,000,000 detection words.
Usage past your credits is billed at the same per-word rate, with no markups, from the same balance as the rest of the Business API, and your dashboard shows every word as it's used. Cancel anytime.
| Usage | Per 1K words | Words included |
|---|---|---|
| AI detection | $0.02 | 5,000,000 |
| Standard humanization | $0.14 | 714,000 |
| Phrasly Ultra | $0.30 | ~333,000 |

AI detection API pricing compared
At $0.02 per 1,000 words, Phrasly has the lowest published per-word rate we found. Because the plan starts at $100 a month, an API with a smaller minimum can cost less at low volumes.
| API | Published price | Per 1,000 words |
|---|---|---|
| Phrasly | $0.02 per 1,000 words; $100/month plan | $0.02 |
| Sapling | $0.005 per 1,000 characters; $5 one-time minimum | ~$0.03 (calculated) |
| Originality.ai | API on the Enterprise plan: $179/month for 1.5M words | ~$0.12 (calculated) |
| Pangram | $0.05 per 100 words | $0.50 |
From each vendor's public pricing page, checked November 9, 2026. "Calculated" means we converted the vendor's own pricing to a per-1,000-word figure. Prices change; check each vendor before you buy.
Limits, languages and models detected
Each request takes 50 words to 15,000 characters of English text and returns a result in under 2 seconds.
| Endpoint | POST https://business.phrasly.ai/api/v1/detect |
|---|---|
| Authentication | x-api-key header |
| Request body | text (string, required) |
| Minimum text length | 50 words |
| Maximum text length | 15,000 characters |
| Response time | Under 2 seconds |
| Supported languages | English (primary) |
| Models detected | ChatGPT, Claude, Gemini and other large language models |
| Output | Label, AI %, Human %, flagged sentences, a 0–1 score for every sentence |
| Rate limits | Generous limits for production use; higher throughput on request |
AI detection API use cases.
Teams use the API wherever they need programmatic AI content detection: student submissions, marketing copy, user-generated content and AI transparency features.
Academic integrity platforms
Screen student essays and assignments for AI-generated content before grading.
Publishing and media
Verify that submitted articles, guest posts and editorial content are human-authored.
Content moderation
Flag AI-generated spam, fake reviews and synthetic comments at scale.
SEO and marketing platforms
Audit content quality so AI-generated copy meets editorial standards before it's published.
Enterprise compliance
Review internal documents, reports and communications for AI authorship disclosure requirements.
SaaS products
Add AI transparency features that show users whether content was created by a human or an AI model.
Use AI detection responsibly
An AI score is a signal for review, not a verdict. Don't auto-reject a candidate or penalize a student based on a score alone.
Use the flagged sentences to start a review, keep a person in the loop for decisions that affect people, and score texts of at least 50 words for the most reliable results.
AI detection API FAQ
Something else on your mind? Email support@phrasly.ai.
What is an AI detection API?
An AI detection API is a web service that checks whether text was written by AI. Your app sends the text in an HTTP request and gets back a score, usually an overall AI percentage and, with some APIs, a score for each sentence, so you can flag AI writing inside your own product.
How fast is the Phrasly AI detector API?
Responses come back in under 2 seconds. Repeated requests for the same sentences are served from a cache, so re-scanning lightly edited drafts is fast.
Which AI models does it detect?
The API detects text generated by large language models such as ChatGPT, Claude and Gemini, as well as other large language models.
Which languages does the API support?
English is the primary supported language.
What are the minimum and maximum text lengths?
Each request needs at least 50 words and can contain up to 15,000 characters. Split longer documents into sections and send each one as its own request.
What does the API return? Do I get sentence-level scores?
Yes. Every response includes a label (AI, Might be AI or Human), AI and human confidence percentages from 0 to 100, the sentences flagged as AI, and a 0–1 AI score for every sentence in the input.
How much does the AI detection API cost?
AI detection costs $0.02 per 1,000 words. The Business plan is $100 a month and includes $100 in API credits, which covers up to 5,000,000 detection words. Usage beyond your credits is billed at the same per-word rate, with no markup.
Can I scan many documents at once?
Yes. Call the API from your own scripts or pipeline within your rate limits, or upload a CSV of up to 100 documents in the dashboard to scan them in one run.
Is there a rate limit?
Business accounts have generous rate limits suitable for production use. Contact Phrasly if you need higher throughput for enterprise workloads.
Can I auto-reject candidates or penalize students based on the score?
No. Treat the score as a signal, not proof. Use the flagged sentences to decide what a person should review, and make the final decision with human judgment.
What's the difference between the API and Phrasly's AI detector?
The API is built for apps and volume: you send text from your own code and get JSON back, including a score for every sentence. The Phrasly AI detector at phrasly.ai/ai-detector is for checking text by hand in your browser.
How do I get started?
Sign up for a business account, subscribe to the plan and generate your API key in the dashboard. You can start making API calls immediately.
Add AI detection to your product.
$0.02 per 1,000 words, a score for every sentence, under 2 seconds.