AI Content Detector API

Our AI Content Detector API offers rapid and accurate detection across 26 languages by using 4 models.

Sample Usage

curl -X POST "https://fapihub.com/v2/detect/" \
  -H "token: API_TOKEN" \
  -F "text=AI text is here" \
  -F "model=fast-detect-basic"

model can be: fast-detect-advanced, fast-detect-basic, glimpse-premium, glimpse-standard

Response

{
    "success": true,
    "status_message": "Succeed",
    "probability": 0.21285641836394117,
    "criticality": -0.316162109375,
    "aipercentage": "21.29%",
    "ntoken": 3,
    "result_text": "Likely human-written"
}

Detection Models

Method	Description
Fast Detect Advanced	Uses larger parameter models (7B) for both sampling and scoring. Offers the highest accuracy among fast detection methods while maintaining good speed. Recommended for general-purpose detection when balance between accuracy and speed is needed.
Fast Detect Basic	Employs smaller parameter models (2.7B) for both sampling and scoring. Faster processing time but slightly lower accuracy. Good for quick initial assessments and situations where computational resources are limited.
Glimpse Premium	Top-tier multilingual detection supporting 26 languages. Uses the most powerful models for scoring text across multiple languages. Highest accuracy but more computationally intensive. Best for critical content verification.
Glimpse Standard	Balanced multilingual detection supporting 26 languages. Uses efficient models to provide good accuracy across languages. Less resource-intensive than Premium while maintaining solid detection capabilities. Suitable for routine multilingual content verification.

Detection Metrics

Metric	Description
Probability	Score between 0-1 indicating likelihood of AI generation. Values above 0.5 suggest AI authorship, while values below 0.5 suggest human authorship.
Criticality	Confidence metric for the detection. Higher values indicate stronger statistical evidence for the classification made by the detector.
Token Count	Number of text chunks analyzed, providing context for detection reliability. Longer texts generally provide more reliable detection results.

How to Interpret Results

• Probability > 0.8: Strong indication of AI-generated content

• Probability 0.6-0.8: Moderate indication of AI-generated content

• Probability 0.4-0.6: Ambiguous, could be either AI or human

• Probability < 0.4: Likely human-written content

For most accurate results, consider using multiple detection methods and comparing their outputs.