Introduction

The 91data API serves cleaned, unified Indian business data — company records sourced from MCA and government registries. It's a read-only REST API: two endpoints, JSON in and out, authenticated with an API key.

Authentication

Every request must include your API key in the X-API-Key header. Requests without a valid key return 401 Unauthorized.

Get a key from your dashboard — Owners and Developers can issue and revoke keys.

curl "https://api.data91.in/companies/U45201AN2022PTC005613" \
  -H "X-API-Key: 91d_your_api_key"

Rate limits

Your plan includes a daily call quota. Going over it does not block requests — calls beyond your quota are still served, but flagged for overage billing. When a response is served over-quota, it carries an X-Quota-Status: over-quota response header. See your Billing tab for your current plan and overage rate.

Quota is counted per HTTP request, not per row — a single /companies/export call streaming a million matching rows counts once against your quota, the same as any other call.

Look up a single company by its Corporate Identification Number (CIN).

curl "https://api.data91.in/companies/U45201AN2022PTC005613" \
  -H "X-API-Key: 91d_your_api_key"

{
  "cin": "U45201AN2022PTC005613",
  "company_name": "INLAND ACE CONSTRUCTION PRIVATE LIMITED",
  "registration_date": "2022-08-25",
  "company_category": "Company limited by shares",
  "company_class": "Private",
  "listing_status": "Unlisted",
  "authorized_capital": "1500000.00",
  "paidup_capital": "1500000.00",
  "company_roc": "ROC Andaman",
  "address": "No 5, Ward No 7, Madhupur Diglipur,Andaman,Andaman Islands,Andaman and Nicobar Islands,744202-India",
  "pin_code": "744202",
  "state": "Andaman and Nicobar Islands",
  "status": "Active",
  "sub_category": "Non-government company",
  "industrial_classification": "Construction",
  "cin_nic_code": "45201",
  "cin_incorporation_year": 2022
}

Returns 404 if no company matches the CIN.

Search companies by name (fuzzy match) and/or exact-match filters, with pagination. The response is wrapped in a results envelope alongside pagination metadata — see Pagination at scale below for how to page through large result sets efficiently.

curl "https://api.data91.in/companies?name=inland&state=Andaman%20and%20Nicobar%20Islands&limit=10" \
  -H "X-API-Key: 91d_your_api_key"

{
  "results": [
    {
      "cin": "U45201AN2022PTC005613",
      "company_name": "INLAND ACE CONSTRUCTION PRIVATE LIMITED",
      "state": "Andaman and Nicobar Islands",
      "status": "Active"
      /* ...remaining fields, same shape as GET /companies/{cin} */
    }
  ],
  "next_cursor": "U45201AN2022PTC005613",
  "has_more": true
}

Pagination at scale

The dataset behind this API is several million rows. offset-based paging works fine for shallow pages or narrow filtered results, but it gets slower the deeper you page — Postgres has to scan and discard every row before the offset on each request. Paging to row 3,000,000 via offset=3000000 means scanning 3 million rows just to skip them.

cursor-based pagination avoids this entirely — it's a direct index seek, so page 50,000 is exactly as fast as page 1, regardless of how deep you go. Take the next_cursor from a response and pass it back as cursor on the next request; stop when has_more is false. Cursor pagination is only available when nameisn't set — name search is ordered by relevance, not CIN, so there's no stable cursor to hand back; use offset there instead.

For pulling large slices of the dataset rather than browsing a few pages, see Bulk export below — it's built for that specifically and is more efficient than paging through thousands of /companies calls.

# First page
curl "https://api.data91.in/companies?state=Maharashtra&limit=100" \
  -H "X-API-Key: 91d_your_api_key"

# Next page — use the previous response's "next_cursor"
curl "https://api.data91.in/companies?state=Maharashtra&limit=100&cursor=U74899MH2021PTC123456" \
  -H "X-API-Key: 91d_your_api_key"

Returns the number of companies matching a filter, without fetching the records themselves — same filter parameters as /companies (no limit/cursor/offset). An unfiltered call returns an instant estimate (Postgres's own table statistics) rather than scanning several million rows to count them — check the exact field in the response to tell which you got. Filtered calls run a real, exact count.

curl "https://api.data91.in/companies/count?state=Maharashtra" \
  -H "X-API-Key: 91d_your_api_key"

{
  "count": 187342,
  "exact": true
}

Streams matching companies as newline-delimited JSON (NDJSON) — one compact JSON object per line, not wrapped in an array. Built for bulk/large-scale consumption: syncing a filtered slice of the dataset into your own system, rather than paging through it a few hundred rows at a time. Same filter parameters as /companies, plus cursor to resume a large export across multiple requests (results are always ordered by CIN — pass the last CIN you received as cursor to continue where you left off, e.g. after a dropped connection).

Process the response as a stream rather than buffering the whole body — see the Python example below. A single export call is capped at 2,000,000 rows; for anything larger, resume with cursor across multiple calls.

curl "https://api.data91.in/companies/export?state=Maharashtra&status=Active" \
  -H "X-API-Key: 91d_your_api_key"

{"cin": "U74899MH2021PTC123456", "company_name": "ACME PRIVATE LIMITED", "state": "Maharashtra", "status": "Active", ...}
{"cin": "U74899MH2021PTC123457", "company_name": "ACME TRADING LLP", "state": "Maharashtra", "status": "Active", ...}
{"cin": "U74899MH2021PTC123458", "company_name": "ACME LOGISTICS PVT LTD", "state": "Maharashtra", "status": "Active", ...}
...one JSON object per line, not wrapped in an array

import requests, json

url = "https://api.data91.in/companies/export"
params = {"state": "Maharashtra", "status": "Active"}
headers = {"X-API-Key": "91d_your_api_key"}

with requests.get(url, params=params, headers=headers, stream=True) as res:
    res.raise_for_status()
    for line in res.iter_lines():
        if not line:
            continue
        company = json.loads(line)
        # process one company at a time — memory usage stays flat
        # regardless of how many millions of rows match
        print(company["cin"], company["company_name"])

Response fields

/companies/{cin}, /companies (inside results), and /companies/export (each NDJSON line) all return the same company object shape.

Errors

For /companies/export, the status code and headers are only meaningful for the initial connection — once streaming starts, a failure partway through ends the connection rather than returning a clean error body. If a stream cuts off unexpectedly, resume with cursor set to the last CIN you successfully received.

More examples

const res = await fetch("https://api.data91.in/companies/U45201AN2022PTC005613", {
  headers: { "X-API-Key": "91d_your_api_key" }
});

if (!res.ok) throw new Error(`91data API error: ${res.status}`);
const company = await res.json();

import requests

res = requests.get(
    "https://api.data91.in/companies/U45201AN2022PTC005613",
    headers={"X-API-Key": "91d_your_api_key"},
)
res.raise_for_status()
company = res.json()