Image OCR

POST https://tools.yukai.uk/api/v1/ocr

Extract text from an image (OCR). Reads all text in a photo, scan or screenshot (receipts, invoices, documents, signs, labels, menus, screenshots; many languages and scripts (20+ tested) including Chinese, Japanese and Korean) with an AI vision model (Llama 4 Scout) instructed to transcribe verbatim: nothing is translated, summarized or corrected. Send an image URL (GET or POST), a base64 image (POST JSON), a file upload (POST multipart/form-data, field image) or the raw image bytes (POST with Content-Type: image/*, options in the query string). Images without text return noText: true. Identical images are served from a 24-hour cache.

Cost

Credits: 27 ($0.0027) per successful call. Failed calls are free. Source price: image-to-text-ocr/image-ocr.

Input

Send a JSON body (POST) or the same fields as a query string (GET).

No field is required.

Field	Type	Default	Description
`url`	string		Public image URL (JPEG, PNG, WebP or GIF, up to 6 MB). Google Drive, Dropbox and OneDrive share links work. Private, local and cloud-metadata addresses are refused on every redirect. Alias: `imageUrl`. For POST you can send `image` instead.
`format`	string	`"text"`	`text` (default): plain text, tables as tab-separated rows. `markdown`: Markdown with headings, lists and tables. `lines`: plain text plus a `lines` array (one string per non-empty line; no coordinates).
`language`	string		Optional hint for the main language of the text, as a name (`Japanese`, `German`, `Traditional Chinese`). Helps with small or stylized script. The text is never translated.
`image`	string		The image itself as base64 or a data URL (`data:image/png;base64,...`), up to 6 MB decoded. POST only. Aliases: `imageBase64`, `base64`.

Examples

Replace tt_live_XXXX with your key from the pricing page.

curl

curl -i -X POST "https://tools.yukai.uk/api/v1/ocr" \
  -H "Authorization: Bearer tt_live_XXXX" \
  -H "Content-Type: application/json" \
  -d '{"url":"https://upload.wikimedia.org/wikipedia/commons/0/08/China_nutrition_facts_label.png","format":"markdown"}'

JavaScript (fetch)

const res = await fetch("https://tools.yukai.uk/api/v1/ocr", {
  method: "POST",
  headers: { "Authorization": "Bearer tt_live_XXXX", "Content-Type": "application/json" },
  body: JSON.stringify({"url":"https://upload.wikimedia.org/wikipedia/commons/0/08/China_nutrition_facts_label.png","format":"markdown"}),
});
console.log(res.status, res.headers.get("x-credits-charged"), res.headers.get("x-credits-balance"));
console.log(await res.json());

Python (requests)

import requests

r = requests.post(
    "https://tools.yukai.uk/api/v1/ocr",
    headers={"Authorization": "Bearer tt_live_XXXX"},
    json={"url": "https://upload.wikimedia.org/wikipedia/commons/0/08/China_nutrition_facts_label.png", "format": "markdown"},
    timeout=120,
)
print(r.status_code, r.headers.get("x-credits-charged"), r.headers.get("x-credits-balance"))
print(r.json())

Response

JSON with "success": true and the result fields; on failure {"success": false, "error": {"code", "message", "retryable"}} and no credits are used. Every answer carries x-credits-charged (credits used) and x-credits-balance (balance after the call); check the balance any time with GET https://tools.yukai.uk/v1/billing/balance.

Full field list and response schema: OpenAPI spec image-ocr (its servers entry is the marketplace gateway; call https://tools.yukai.uk with your key).

← All routes · Pricing