Question 1

What types of AI-generated images can this API detect?

Accepted Answer

Our API detects images from all major AI generators including Midjourney (v4–v6), DALL-E 2 & 3, Stable Diffusion (1.5, XL, 3.0), Flux, Adobe Firefly, Google Imagen, and 15+ other models. We also detect hybrid images that combine AI generation with manual editing.

Question 2

How accurate is the AI Image Detector API?

Accepted Answer

Our detection model achieves 99.1% accuracy on our benchmark dataset of 10M images. We maintain separate precision and recall metrics per generator: 99.4% for Midjourney, 98.9% for DALL-E, 99.2% for Stable Diffusion, and 98.7% for Flux. We continuously retrain as new models emerge.

Question 3

Can the API detect partially AI-edited images?

Accepted Answer

Yes. Beyond fully synthetic images, our API detects AI inpainting, outpainting, face swaps, background replacements, and generative fill edits. The pixel-level heatmap response highlights exactly which regions were AI-modified.

Question 4

How does source model attribution work?

Accepted Answer

Our classifier analyzes frequency-domain signatures, noise patterns, and structural artifacts unique to each generator. It returns the most likely source model with a confidence score. When multiple generators were used (e.g., Stable Diffusion base + Midjourney upscale), it flags the image as multi-model.

Question 5

Is there a free tier?

Accepted Answer

Yes. The free tier includes 500 image scans per month with full API access, model attribution, and heatmap generation. No credit card required.

AI Image Detector API Documentation

Quick Start Guide

API Endpoints

Authentication

Rate Limits

SDKs & Libraries

Error Handling