Models / Available
Orca
Orca estimates location from visible evidence in an image. Use it through the app for interactive review or the REST API for structured responses.
Supported inputs
The REST endpoint accepts a publicly fetchable image_url or base64 image data in image_b64. Supply one source image per geolocation request. See the endpoint reference for payload limits and supported request fields.
Analysis settings
- D1: a broad location estimate.
- D2: additional explanation and visual clues.
- D3: deeper evidence review, with forensic and calibration fields when available.
Depth controls the work requested, not guaranteed precision. The same model identifier, orca-1.5, is used for all three.
Read the output
Location, confidence, reasoning, and model information describe the returned estimate. API confidence is 0–1. At D2 and above, visual_clues contains objects with label and detail; detail can be empty. Optional fields may be absent. A saved analysis_id can be null when history is not retained.
Limitations
Generic interiors, repeated architecture, blurred signs, altered images, and missing context can make a location ambiguous. Coordinates can describe a broad candidate area. A precise-looking pin or a high confidence score is not independent verification. Check the image, compare alternatives, and preserve unresolved questions.