Firecrawl map endpoint limit 5000 vs Sume crawl_map limit 100

Firecrawl's map endpoint has a limit default of 5000 and a maximum of 100000. Sume's map (POST /v1/firecrawl/map, tool crawl_map) takes limit from 1 to 100, default 100. It is built for locating pages, not for dumping a whole site's URL list.

How do the limits compare?

Both take a URL and a limit; the ceiling is what differs.

What is crawl_map for?

The tool description says: locate pages within a known site before reading them. It returns links with titles and descriptions. Its next step is to pick relevant URLs, then call crawl_scrape; it also says never to crawl merely to locate one page.

What does the request look like?

Map once, filter the returned titles in your own code, then scrape only the winners. The shape of the response is not specified in the OpenAPI schema, so inspect data before relying on a field name.

const headers = {
  Authorization: "Bearer " + process.env.SUME_API_KEY,
  "Content-Type": "application/json",
};
const post = (path, body) =>
  fetch("https://api.sume.com/v1/firecrawl/" + path, {
    method: "POST",
    headers,
    body: JSON.stringify(body),
  }).then((r) => r.json());

const map = await post("map", { url: "https://example.com", limit: 100 });
console.log(JSON.stringify(map.data, null, 2).slice(0, 2000));
// choose a URL from the printed links, then:
// await post("scrape", { url: chosenUrl, timeout_ms: 30000 });

What if I need every URL on a large site?

A map call returns at most 100 links, so it will not give a complete inventory of a large site. For content across a site section the docs point to crawl_site, which is asynchronous; see MCP tools and gates. Use the map to find sections and scrape the pages you need.