Last updated October 5, 2026
Privacy policy
Model-file processing is local, but visiting the website still creates ordinary network traffic. This page explains the distinction.
Model files and parsed results
When you select a GGUF or ONNX file, browser JavaScript reads it on your device. The application does not contain a model-upload request or an AI Model Inspector API, and the model contents are not sent to an AI Model Inspector server. Parsed metadata and graph information stay in the current page unless you use your browser to download the JSON report.
The app processes only files you choose through the file picker or drop area. GGUF parsing reads chunks as needed for the header, metadata, and tensor descriptors. ONNX parsing loads the selected ONNX file into browser memory. Closing or reloading the page clears the in-page results.
SHA-256 hashing
Hashing is optional. When requested, a web worker reads the file locally in 8 MiB chunks and calculates SHA-256 using the bundled hash-wasm library. The displayed hash is not sent by application code. A hash is a fingerprint and may still be sensitive when it can identify a known file, so share it only when appropriate.
Website and hosting traffic
The site is delivered through Firebase Hosting. As with most websites, requests for pages and assets disclose network information such as your IP address, browser headers, requested URL, and timing to the hosting provider and network intermediaries. This ordinary traffic is separate from reading your selected model file.
The application does not include Google Analytics, account sign-in, a comment system, or its own tracking API. It saves one preference, ami-theme, in browser local storage when you change the color theme. That value records only the chosen theme on your device.
Google advertising
The inspector page loads the Google AdSense script from pagead2.googlesyndication.com. Loading it makes a request to Google. Google and its partners may receive device and network information and may use cookies, local storage, or similar technologies to provide, measure, and protect advertising, subject to consent and regional requirements.
Advertising requests do not receive the model file from AI Model Inspector's code. However, an advertising script runs on the same webpage as the inspector, so users with stricter confidentiality requirements may prefer to block third-party scripts or run the open-source project locally.
Learn how Google handles data in its privacy policy and its explanation of data used from partner sites. You can review advertising controls at My Ad Center.
External links
Links to GitHub and Buy Me a Coffee take you to those services. If you follow them, the destination receives an ordinary page request and applies its own privacy policy. AI Model Inspector does not receive account or payment details from those services through this static application.
Retention, rights, and changes
AI Model Inspector has no application database for model files, reports, hashes, or user accounts. Hosting and advertising providers may retain request or advertising data under their own policies. Requests concerning data controlled by those providers should be directed to the relevant provider.
This policy may change when the site's code or third-party services change. Material changes will be reflected here with a new update date. Questions or privacy-related bug reports can be filed through the project's public issue tracker; do not include private model data in an issue.