API Documentation - Qwip AI Detection

Base URL

https://api.qwip.io

Environment: Production (public beta)

Protocol: HTTPS only (TLS 1.3)

Rate Limit: 100 requests per minute per IP address

Authentication

No authentication required. The API is publicly accessible for reading and writing. Rate limiting is based on IP address to prevent abuse.

Note: Future versions may introduce optional API keys for higher rate limits or advanced features. For now, the API is completely open.

Response Format

All responses are in JSON format with UTF-8 encoding.

Success Response

{
  "success": true,
  "data": { // ... endpoint-specific data }
}
            

Error Response

{
  "success": false,
  "error": "Error message"
}
            

Endpoints

POST /api/check

Query the community database to see if an image (identified by its BLAKE3 hash) has been analyzed before. Returns aggregated detection results from community contributions.

Request Body

Parameter	Type	Required	Description
blake3Hash	string	Required	64-character hexadecimal BLAKE3 content hash of the image

Example Request

POST /api/check
Content-Type: application/json

{
  "blake3Hash": "a7f3d8c9e2b4f1a6d5c3e8b7a2f9d4c1b6e3a8f7d2c9b5e4a1f8d3c7b2e6a9f4"
}
                

Response (Found)

{
  "found": true,
  "data": {
    "likelyAI": true,
    "confidence": 0.87,
    "modelUsed": "cifake",
    "voteCount": 5,
    "firstSeen": "2025-12-15T10:30:00Z",
    "lastSeen": "2025-12-27T14:22:00Z"
  }
}
                

Response (Not Found)

{
  "found": false
}
                

Response Fields

Field	Type	Description
found	boolean	Whether the hash exists in the database
likelyAI	boolean	Community consensus: true if likely AI-generated
confidence	number	Aggregated confidence score (0.0 - 1.0)
modelUsed	string	Most common model used for detection
voteCount	number	Number of independent analyses contributed
firstSeen	string	ISO 8601 timestamp of first contribution
lastSeen	string	ISO 8601 timestamp of most recent contribution

POST /api/contribute

Contribute a detection result to the community database. This helps build a shared knowledge base of AI-generated images. All contributions are anonymous.

Request Body

Parameter	Type	Required	Description
blake3Hash	string	Required	64-character hex BLAKE3 hash
hashVector	object	Required	5 perceptual hashes (see below)
likelyAI	boolean	Required	Detection result (true = AI)
confidence	number	Required	Model confidence (0.0 - 1.0)
modelUsed	string	Required	"cifake", "genimage", "mobilevit", or "unknown"

Hash Vector Object

Field	Type	Description
mean	string	Mean perceptual hash (64-bit integer as string)
gradient	string	Gradient perceptual hash
doubleGradient	string	Double gradient perceptual hash
block	string	Block median perceptual hash
dct	string	DCT perceptual hash

Example Request

POST /api/contribute
Content-Type: application/json

{
  "blake3Hash": "a7f3d8c9e2b4f1a6d5c3e8b7a2f9d4c1b6e3a8f7d2c9b5e4a1f8d3c7b2e6a9f4",
  "hashVector": {
    "mean": "18379468920823898112",
    "gradient": "18015498021093556224",
    "doubleGradient": "18374389475892961280",
    "block": "18302628773641904128",
    "dct": "18374673854875439104"
  },
  "likelyAI": true,
  "confidence": 0.87,
  "modelUsed": "cifake"
}
                

Response

{
  "success": true,
  "isNew": false,
  "voteCount": 6,
  "message": "Contribution recorded, aggregated with 5 existing votes"
}
                

Response Fields

Field	Type	Description
success	boolean	Whether contribution was recorded
isNew	boolean	True if this is the first analysis of this image
voteCount	number	Total number of analyses (including this one)
message	string	Human-readable confirmation message

Error Codes

HTTP Status	Meaning	Common Causes
200	Success	Request completed successfully
400	Bad Request	Invalid hash format, missing required fields, invalid model name
429	Rate Limited	Exceeded 100 requests/minute from your IP
500	Server Error	Database connection failure or internal error

Vote Aggregation Algorithm

When multiple users analyze the same image, their results are combined using a weighted average:

// New confidence score
new_confidence = (old_confidence × vote_count + contribution_confidence)
                 ÷ (vote_count + 1)

// Increment vote counter
vote_count = vote_count + 1

// Update consensus (majority vote)
likely_ai = (ai_votes > real_votes)
            

This ensures that the confidence score reflects the average opinion of all contributors, giving equal weight to each analysis.

Code Examples

JavaScript (Fetch API)

// Query for existing result
const response = await fetch('https://api.qwip.io/api/check', {
  method: 'POST',
  headers: { 'Content-Type': 'application/json' },
  body: JSON.stringify({
    blake3Hash: 'a7f3d8c9e2b4f1a6...'
  })
});

const data = await response.json();
if (data.found) {
  console.log(`AI probability: ${data.data.confidence}`);
}
            

Python (Requests)

import requests

# Contribute detection result
response = requests.post(
    'https://api.qwip.io/api/contribute',
    json={
        'blake3Hash': 'a7f3d8c9e2b4f1a6...',
        'hashVector': {
            'mean': '18379468920823898112',
            'gradient': '18015498021093556224',
            'doubleGradient': '18374389475892961280',
            'block': '18302628773641904128',
            'dct': '18374673854875439104'
        },
        'likelyAI': True,
        'confidence': 0.87,
        'modelUsed': 'cifake'
    }
)

print(response.json())
            

cURL

curl -X POST https://api.qwip.io/api/check \
  -H "Content-Type: application/json" \
  -d '{"blake3Hash":"a7f3d8c9e2b4f1a6..."}'
            

Important Notes:

The API is public and rate-limited by IP (100 req/min)
No authentication means anyone can query and contribute
Contributed data is permanent and cannot be deleted easily
Do not contribute personal or sensitive image hashes

Need Help or Have Questions?

We're here to help you integrate with the Qwip API.

Full Documentation → Report Issues → Contact Support →