Verification-grade geolocation with safety built into the product, not bolted on.
Raven serves verified government agencies. Oceanir serves everyone whose job is verifying images: journalists, legal teams, insurance SIU, OSINT researchers. Safety is part of the architecture. Oceanir reads the scene itself (landmarks, signage, terrain, architecture) and never faces. No facial recognition, no people search, no tools for tracking private individuals, and your uploads are not used to train models unless you explicitly opt in. Forensic-grade evidence output, with the guardrails written into our terms rather than left to trust. Free to start.
Oceanir vs Raven (formerly GeoSpy)
Comparison reflects publicly available information at time of writing.
Built for verification, not just guessing a pin
Scene analysis, never people analysis
Oceanir geolocates from what a place looks like: landmarks, signage, terrain, architecture. It does not do facial recognition, biometric identification, person matching, or people search, and our terms prohibit using it to locate or track private individuals. It answers where an image was taken, not who is in it.
Transparent and accountable
Every analysis returns ranked location candidates with confidence scores and a visible evidence trail, so a human reviewer can see how certain a result is and what the alternatives are before acting on it. Defensible enough for a case file, honest enough to show you when it is unsure.
Your images stay yours
By default, uploads are never used to train Oceanir models, and our inference providers are contractually barred from training on them too. We keep strict internal protocols around how imagery is processed, and the full policy is public in our terms and privacy pages, not a promise in a sales deck.
Raven (formerly GeoSpy) alternative questions
Yes. Graylark says GeoSpy became Raven in April 2026, built by the same team on the same visual-geolocation foundation, for qualified law-enforcement, government, public-safety, intelligence and enterprise organizations. The free public GeoSpy platform had already ended in early 2025.
Oceanir is open to journalists, investigators, insurance SIU teams, legal researchers and anyone else who verifies images. You sign up, upload a photo and get ranked candidate locations with confidence scores. There is no agency verification and no sales call, and the free scan needs no card.
Raven is offered to qualified organizations with no public pricing. Oceanir has a free scan open to anyone, and paid plans with public prices add street-level checks and PDF evidence bundles.
Yes. Raven is the platform GeoSpy became, built for verified law enforcement and intelligence agencies. Oceanir is openly available to journalists, investigators, insurance SIU teams, and legal researchers, with no agency verification required and safety commitments written into our terms.
Safety is designed into the product. Oceanir analyzes scenes, not people: no facial recognition, no biometric identification, no person matching, no people search. Our terms prohibit using the platform to locate or track private individuals. Uploads are not used to train models unless you explicitly opt in, and our inference providers are barred from training on your images. Every result ships with confidence scores and a visible evidence trail so a human makes the final call.
Ranked location candidates with confidence scores, Street View verification, D3 forensic depth with multi-pass zoom, PDF evidence bundles for case files, and a Chrome extension for right-click image analysis. No agency account required. Pro access via standard subscription.
Accuracy depends on what the photo shows. D3 forensic depth adds Street View verification and keypoint matching that improves precision on images with strong visual cues.
Yes. Oceanir offers a full REST API and MCP server for reviewed integrations. Both x-api-key and Authorization: Bearer headers are supported, with limits scoped to the evaluation or production contract.
It helps to separate GeoSpy's two figures. Its base global model is documented at 1 to 50 km accuracy on Graylark's own site. Its meter-level claim applies only to SuperBolt, a separate scene-matching system that works inside the specific cities Graylark has pre-indexed as custom models. For context, the strongest published academic geolocation models still place only about 11 to 20 percent of images within 1 km worldwide, so true street-level accuracy everywhere is not something any tool delivers today. Oceanir's difference is not a bigger headline number, it is calibrated confidence and visible reasoning on every result, so you see when it has an exact address versus only a neighborhood instead of a confident wrong pin. The honest way to compare is to run both on your own images.
Yes. Start free and run your own images with known locations, including the hard ones (non-Western, rural, interiors), and compare the results and the confidence scores side by side. We would rather you verify it on your own cases than take a benchmark on faith.