Captcha types
Five recognition types. Each has its own endpoint under /v1/recognition/, and each also works on the legacy root path with a type field.
task (the instruction text, for example "bus"), grid ("3x3", "4x4" or "1x1") and image_data. For 3x3 and 4x4, send one image of the whole grid. For 1x1, send 1 to 9 tile images.
One boolean per tile. true means: select this tile. Tiles run left to right, top to bottom.
{"data": [false, true, false, true, true, false, false, false, false]}Exactly one image in image_data. No task text.
An array with one string: the recognised text. The text is case-sensitive.
{"data": ["K7m2Q"]}task (the instruction text) and image_data: one complete challenge screenshot, or two images in this order: background, then the transparent puzzle piece.
An object with x, the offset as a percentage of the background image width.
{"data": {"x": 42.5}}task (for example "Pick the sheep") and one image with the 3x2 grid of tiles.
One boolean per tile in row-major order. true marks the tile to pick.
{"data": [false, false, false, true, false, false]}task (for example "Match the number of rocks with the number on the left") and one uncropped game image.
One boolean per answer tile.
{"data": [false, true, false]}Images go in image_data as base64 strings in an array. The request body is limited to 4.5 MB. JSON and url-encoded form bodies are supported, multipart is not. Full formats and examples are in the docs.
Create an account, get your key and change the base URL. Paid plans open soon. We e-mail you when checkout is live.