The Exa comparison rested on one homepage claim, then cross-checked broader first-party scale statements
I reviewed Exa’s homepage and three Exa blog posts about its index and infrastructure. The earlier comparison used the homepage’s then-visible “80 billion documents served” figure; the live first-party material now reports 100 billion documents, so the earlier 80B comparison is no longer current.
100B
current documents claim
1.4T
current globally sourced URLs
500B+
URLs tracked, May 2026
4
first-party pages reviewed
What each Exa page contributed
blue = quantified first-party claim
Dec 2024
Billions of vectors searchable in under 100 msInfrastructure context, not a document-count figure. exa.ai
May 2026
Over 500 billion URLs tracked by Exa’s crawlersA crawler-coverage claim; not equivalent to served documents. exa.ai
Homepage snapshot
80 billion documents served in its vector DBThis was the figure used in the earlier comparison. A later homepage variant returned 100 billion, showing that this figure changed. exa.ai
Aug 2026
1.4 trillion URLs and 100 billion documentsThe clearest current first-party description: Exa calls this a “globally-sourced index.” exa.ai
What the review shows
1
The 80B number was directly observed, not inferred. Exa’s homepage text said “80 BILLION” and “documents served in our vector DB, the world’s largest independent index.”
2
Exa’s disclosed document count has since risen to 100B. Its August 17 Firefox announcement explicitly says the index contains 1.4T URLs and 100B documents. exa.ai
3
URLs and documents were kept separate. The 500B tracked-URL and 1.4T sourced-URL figures describe broader URL universes, not necessarily retrievable documents.
4
The earlier ranking needs updating. On current disclosed counts, Exa and Keenable both report 100B documents; the public figures do not establish which is larger.
| Page | Scale found | What it counts | How it was used |
|---|---|---|---|
| Homepage | 80B at earlier snapshot; 100B in another variant | Documents served in vector DB | Primary source for the earlier 80B comparison |
| Firefox announcement | 1.4T URLs; 100B documents | Globally sourced index | Best current first-party count |
| Series C announcement | 500B+ URLs | URLs tracked by crawlers | Excluded from document ranking |
| Vector DB engineering | Billions of vectors | Searchable vectors | Technical context only |
Method: live web review on 26 Aug 2026 of 4 first-party Exa pages, supplemented by search-result snapshots. “Documents” is reported publisher wording, not an independently audited count; URL, document and vector metrics are not treated as interchangeable.