Common Crawl index API: Jul-Sep 2026 graph and crawl IDs

crawlgraph now serves Common Crawl's Jul-Sep 2026 graph, identified by cc-main-2026-jul-aug-sep. it is built from the CC-MAIN-2026-30, CC-MAIN-2026-34, and CC-MAIN-2026-39 crawls and covers 133.2M domains and 2.15B domain-level links. lookups use it by default; pass the id as release_id to pin it:

crawlgraph's newest queryable domain graph is cc-main-2026-jul-aug-sep. Common Crawl's newest published web graph is cc-main-2026-aug-sep-oct; it is not yet queryable on crawlgraph. Check Common Crawl's web graph release list for the upstream source. Published and queryable are different states.

bash

curl -X POST https://crawlgraph.com/api/v1/backlinks \
  -H "Authorization: Bearer cg_live_…" \
  -H "Content-Type: application/json" \
  -d '{"domain":"example.com","release_id":"cc-main-2026-jul-aug-sep"}'

fewer links upstream, steady profiles

the headline change is size. Common Crawl's domain graph grew in domains and shrank in links:

text

release                     domains        domain-level links
cc-main-2026-apr-may-jun    121.1M         3.90B
cc-main-2026-jul-aug-sep    133.2M         2.15B
change                      +10%           -45%

both rows are Common Crawl's own published totals, not crawlgraph estimates. the link total has fallen in every monthly graph since Feb-Apr 2026 (4.76B), and Common Crawl's release notes do not give a reason.

the drop did not show up where we looked. we counted referring domains for fourteen established sites in both releases, and every one stayed within -1% and +15%. what did move is membership: 22% to 37% of each site's Apr-Jun referring domains were not seen again in Jul-Sep, and at least as many new ones replaced them.

text

referring domains        apr-jun   jul-sep   change   not seen again
wikipedia.org          2,180,577 2,198,837     +1%        23%
github.com               751,171   766,286     +2%        23%
nytimes.com              433,432   460,992     +6%        25%
bbc.com                  257,756   263,062     +2%        29%
ahrefs.com               112,015   121,656     +9%        22%
moz.com                   50,389    49,948     -1%        29%
asana.com                 38,408    40,106     +4%        30%
backlinko.com             27,798    27,996     +1%        29%
smashingmagazine.com      19,155    19,032     -1%        32%
basecamp.com              11,750    11,967     +2%        32%
paulgraham.com            10,127    10,467     +3%        31%
kottke.org                 5,225     5,248      0%        32%
indiehackers.com           4,993     5,748    +15%        36%
danluu.com                 1,283     1,453    +13%        37%

fourteen sites are a sample, not the whole graph, so a smaller or newer site can still move more than these did. the counts come from the two release databases crawlgraph serves.

fewer observed links is not lost links

a domain missing from this release was not observed linking in these three crawls. that is different from proving the link was removed. before you chase a missing link, run the check in a missing backlink is not a lost backlink.

the Apr-Jun 2026 release stays queryable. pin it when you need numbers that match an earlier report:

bash

curl -X POST https://crawlgraph.com/api/v1/backlinks \
  -H "Authorization: Bearer cg_live_…" \
  -H "Content-Type: application/json" \
  -d '{"domain":"example.com","release_id":"cc-main-2026-apr-may-jun"}'

the ids used by this release

two different identifiers get called "the common crawl index", and mixing them up is the usual reason an api call 404s:

crawl ids like CC-MAIN-2026-39 name a single crawl. the pattern is CC-MAIN-<year>-<week>. the cdx endpoint at index.commoncrawl.org accepts these, always with the -index suffix and a url= parameter.
graph release ids like cc-main-2026-jul-aug-sep name a hyperlink-graph release built from three crawls. crawlgraph queries these, and release_id expects one.

for Common Crawl's per-crawl url index instead of the link graph, the cdx endpoint takes one crawl id at a time:

text

https://index.commoncrawl.org/CC-MAIN-2026-39-index?url=example.com&output=json

Common Crawl also publishes overlapping monthly graphs. the two between Apr-Jun and Jul-Sep, cc-main-2026-may-jun-jul and cc-main-2026-jun-jul-aug, share crawls with both, so crawlgraph skips them and moves one full quarter at a time. that keeps each comparison between two releases with no source crawl in common.

how to check which graph is newest

do not hardcode an id from a blog post, this one included. Common Crawl lists every published web graph, newest first, with its source crawls and sizes:

bash

curl https://index.commoncrawl.org/graphinfo.json

crawlgraph lists the releases it has indexed, with an available flag so you know which ones answer queries:

bash

curl https://crawlgraph.com/api/v1/releases \
  -H "Authorization: Bearer cg_live_…"

the two lists run at different speeds. Common Crawl published this graph on 22 September 2026, and crawlgraph ingests and verifies a graph before serving it. the release schedule explains that lag.

compare Apr-Jun with Jul-Sep

GET /api/v1/changes compares one domain across two releases. with the pair spelled out:

bash

curl "https://crawlgraph.com/api/v1/changes?domain=example.com&from=cc-main-2026-apr-may-jun&to=cc-main-2026-jul-aug-sep" \
  -H "Authorization: Bearer cg_live_…"

the response separates added and removed observations and reports authority movement. for the sites in the sample above, 22% to 37% of referring domains changed in each direction, so a long removed list is normal; read it as a snapshot difference. request and response details are in the changes api reference. the MCP server's releases and backlink_changes tools return the same data to an AI assistant.

run a lookup on the new graph

no key yet? enter a domain below to see which referring domains the Jul-Sep 2026 graph observed for it.

livetry it on your own site

run this query against your domain - free

first 5 backlinks free. no signup required.

faq

why not the Aug-Oct 2026 graph?

cc-main-2026-aug-sep-oct was published on 22 October 2026 and shares two crawls with Jul-Sep. crawlgraph moves a full quarter at a time, so the next release it plans to index is the one built from the Oct-Dec 2026 crawls.

how do i pin a query to a specific release?

send release_id in the body of POST /api/v1/backlinks. omit it and crawlgraph uses the latest queryable release, which is right for monitoring and wrong for a reproducible study.

where are the Apr-Jun 2026 notes?

in the Apr-Jun 2026 release notes. that release remains queryable by id.

do i need a key to query the new graph?

not for the site. the lookup box above runs on the current release with no signup. the api has a free tier with a self-serve key, documented in the api reference.