GeezerVison

POST

/v2/parse

import cohere
co = cohere.ClientV2()
response = co.parse(
    model="parse-v5.0",
    document={
        "type": "image_url",
        "image_url": "https://cohere.com/favicon-32x32.png",
    },
    output_format="markdown",
)
print(response)

{
  "id": "8f2a1c3e-4b5d-6e7f-8091-a2b3c4d5e6f7",
  "pages": [
    {
      "type": "markdown",
      "index": 0,
      "markdown": {
        "content": "# Sample Document\n\nCohere builds AI that understands language.\n\n![Company logo](img-0)\n",
        "images": [
          {
            "id": "img-0",
            "description": "Company logo",
            "category": "logo",
            "bounding_box": {
              "top_left_x": 12,
              "top_left_y": 8,
              "bottom_right_x": 48,
              "bottom_right_y": 44
            },
            "bounding_box_normalized": {
              "top_left_x": 0.04,
              "top_left_y": 0.03,
              "bottom_right_x": 0.15,
              "bottom_right_y": 0.14
            }
          }
        ]
      }
    }
  ],
  "meta": {
    "api_version": {
      "version": "2"
    },
    "billed_units": {
      "pages": 1
    }
  }
}

Parse a document image into structured output. Use output_format to select blocks or markdown (default).

Currently supports document.type = image_url only (data URI or remote http(s) image URL). PDF / file URL inputs are not yet supported.

Image limits: 20 MB file size; 50 megapixels or 200 MB decoded (whichever is exceeded first).

Authentication

AuthorizationBearer

Bearer authentication of the form Bearer <token>, where token is your auth token.

Headers

X-Client-NamestringOptional

The name of the project that is making the request.

Request

This endpoint expects an object.

modelstringRequired

The name of a compatible Cohere parse model.

documentobjectRequired

Document to parse. Currently only image_url documents are supported.

output_formatenumOptionalDefaults to markdown

Selects the page payload shape in the response. Defaults to "markdown".

"markdown": each page includes markdown content. Tables are inlined as HTML, and images are referenced as ![<description>](<image_id>) with matching entries in page.markdown.images.
"blocks": each page includes an ordered blocks array of text, image, and table content regions, including bounding boxes where available.

Allowed values:

Response

idstring

Unique identifier for the parse response.

pageslist of objects

Parsed pages in document order.

metaobjectOptional

finish_reasonenumOptional

The reason parsing finished. For Parse, this is only ever COMPLETE or MAX_TOKENS; MAX_TOKENS indicates the output was truncated because it exceeded the maximum output token limit.

Errors

400

Bad Request Error

401

Unauthorized Error

403

Forbidden Error

404

Not Found Error

422

Unprocessable Entity Error

429

Too Many Requests Error

498

Invalid Token Error

499

Client Closed Request Error

500

Internal Server Error

501

Not Implemented Error

503

Service Unavailable Error

504

Gateway Timeout Error