POST
/
parse
/
parsepdf
curl --request POST \
  --url https://api.sarvam.ai/parse/parsepdf \
  --header 'Content-Type: multipart/form-data' \
  --form page_number=1 \
  --form sarvam_mode=large \
  --form prompt_caching=false
{
  "output": "<string>"
}

Headers

​
api-subscription-key
string
default:

Your unique subscription key for authenticating requests to the Sarvam AI Speech-to-Text API. " "Here are the steps to get your api key", example="dxxxxxx-bxxx-4xxx-axxx-cxxxxxxxxxxx

Body

multipart/form-data
​
pdf
file
required

Upload the PDF file you want to parse. This should be uploaded as a form input if you're using multipart/form-data Note: Sarvam Parse supports only English PDFs currently.

​
page_number
string
default:1

The page number you want to extract data from. This is a one-based index (meaning, the first page is 1).

​
sarvam_mode
enum<string>

The mode of parsing to use:

  • small: Use this mode for economical and fast parsing
  • large: Use this mode for highest precision parsing
Available options:
small,
large
​
prompt_caching
enum<string>

Whether to cache the prompt for the parse request. This is useful when running multiple requests to the parsing endpoint.

Available options:
true,
false

Response

200
application/json
Successful Response
​
output
string | null

The base64 encoded HTML string corresponding to the parsed page. The output will be an empty string if parsing fails for some reason.