Changelog

A basic score tells you whether an image looks generated. Deep Scan tells you why, and where. It is available today, along with the plans that pay for it.

Deep Scan, Powered by TruthScan

Run it on any finished scan to add a forensic layer from TruthScan, which analyses generative artifacts, pixel-level manipulation, compression inconsistencies and metadata anomalies.

• Region heatmap: see exactly which parts of the image carry generated or manipulated signals, instead of a single number for the whole picture
• Detector signals: four independent checks reported side by side — the ML model verdict, SynthID watermark, metadata and OCR — each with its own confidence
• Agreement rating: how strongly those detectors agree, reported as weak, moderate or strong
• Plain-language reasoning: key indicators, visual patterns and recommendations written out, not just scores
• Content tags: what the image actually depicts
• Warnings: conditions that weaken the analysis, such as heavy compression

Two-Stage Pipeline

Basic and deep analysis are now separate stages on one task.

• Basic runs first and settles on its own; deep is opt-in per image
• Live progress through each step: fetching, analyzing, aggregating, notifying
• The combined score weights TruthScan at 60% and Illuminarty at 40%
• A deep analysis that fails or is stopped refunds its credits and leaves the basic result intact

Pricing

• Free: 10 guest scans per UTC day, plus 10 credits on sign-up
• Pro — $19.90/month or $199/year: 2,000 credits every month
• Credit packs: 5,000 credits for $49, or 10,000 for $99, one-time add-ons that stack on Pro
• Enterprise: custom volume — talk to us at support@pixelsniff.com

Deep Scan costs 5 credits; a basic scan stays at 1.

Operations

• Admin console for detection channels, per-channel daily and monthly limits, and queue concurrency
• Capacity-aware queueing: work is held rather than dropped when a channel is at its limit
• Email reports covering both stages

Deep Scan reports forensic signals and confidence, not certainty. Results are for reference only and should not be used as the sole basis for a decision about a person, a claim or a piece of evidence.

Today PixelSniff is live. Upload a photo and find out whether it was made with AI — free, with no account required.

Basic Scan, Powered by Illuminarty

Every upload runs through Illuminarty, a detection engine that combines several computer-vision algorithms to estimate how likely an image came from a public AI generation model.

• AI likelihood score: one clear 0–100 result instead of a vague verdict
• Plain verdicts: scores at or above 70 read as AI, at or below 30 as human, and anything between is reported as uncertain rather than forced either way
• Broad generator coverage: traces left by Midjourney, DALL·E, Stable Diffusion, Flux, Gemini and other popular generators
• Deepfakes and AI edits: synthetic faces, face swaps, inpainted and retouched regions

Free for Everyone

• No sign-up to try: guests get 10 basic scans per UTC day
• 10 free credits on every new account, valid for one month
• 1 credit per basic scan once you are signed in

Uploads

• Formats: JPG, PNG, WebP, GIF, AVIF, BMP, TIFF and HEIC/HEIF
• Size: 1KB to 10MB per image
• Queue-aware: when demand is high your scan waits in line; leave an email and the result is sent to you as soon as it is ready

Accounts and History

• Email and Google sign-in
• Scan history with the score and verdict for every image you have checked
• Email reports for finished scans

Under the Hood

• Built on TanStack Start and React 19, deployed as a single Cloudflare Worker
• Images are used only to run the check and are never shared or published
• English and Chinese throughout

Detection results are probabilistic and provided for reference only. No detector is perfect — treat a score as evidence, not proof.