FAQ
Common questions about GeoInfer image geolocation, accuracy, pricing, privacy and supported images.
Answers to the questions we get most. For API errors such as 401 or 429, see Support.
FAQ
What does GeoInfer do?
GeoInfer estimates where a photo was taken from what is visible in it. It returns ranked candidate locations with GPS coordinates, place names and a confidence radius.
Does GeoInfer need GPS or EXIF metadata?
No. It is one trained model that reads the pixels. It ignores EXIF, so it works on screenshots, re-uploads and photos with the metadata removed.
Does GeoInfer search the web or use reverse image search?
No. It does not search the web, run a reverse image search or call agents. The image itself is the only input, so it works on photos that have never been posted online.
How accurate is GeoInfer?
On the IM2GPS3k benchmark the global model is within 1 km 50% of the time, within 25 km 61.20%, within 200 km 78.93%, and gets the country right 93.3% of the time. High-accuracy regional models reach 97% within 1 km in the areas they cover.
Does GeoInfer identify people or read text?
No. It does not do facial recognition and does not read signs or text. It uses architecture, terrain, vegetation, roads and infrastructure.
How much does it cost?
A global prediction costs 1 credit and a high-accuracy prediction costs 3. Top-ups start at $4.99 for 50 credits, and monthly plans are listed at geoinfer.com/en/pricing.
Which image formats are supported?
The API accepts JPEG, PNG, GIF, BMP, TIFF, WebP, HEIC and HEIF up to 10 MB. The Telegram bot accepts JPEG, PNG and WebP.
Can I use GeoInfer without writing code?
Yes. Use the web app at app.geoinfer.com, the Telegram bot @geoinferbot, the Chrome extension or the Maltego transform. All of them share the same account and credits.
Is there a free trial?
Yes. The web app has a free tier, and every new Telegram chat with @geoinferbot gets 1 free credit.
Can GeoInfer run on-premise?
Yes, for vetted institutions. Contact support@geoinfer.com about on-premise and air-gapped deployments.