Exa

Neural + keyword index T1 Search /exa

Exa is a search engine that retrieves by meaning rather than keywords, using neural embeddings to surface pages that match the intent behind a query. Beyond semantic search it offers a ‘find similar’ capability, clean first-party content extraction, and a built-in answer-and-research layer — so you can go from a fuzzy idea to relevant sources to a synthesized response in one place. It’s the strongest fit for discovery, research and recommendation use cases where keyword search falls short, and it’s certified for sensitive domains like healthcare.

Entry price
$7 / 1k (+content $1/1k)
MCP
official
Latency

Choose if

You want semantic discovery, 'find similar', research, or a built-in /answer endpoint.

Avoid if

You need full keyword/SERP parity — content and summary bill per page per type and stack.

Evals

How Exa performed on the curator-fleet web-access evals — the real measured numbers behind each composite, ranked best-first.

Answercomposite 7.9/10 Top pick#1 of 7
1.00
Correctness
0.48
Citation faithfulness
1.5 s
Latency p50
$0.005 / call
Cost
0%
Errors
Best citation faithfulness.
Searchcomposite 6.7/10 #6 of 11
0.61
Recall@10
1.6 s
Latency p50
$0.014 / useful result
Cost
0%
Errors
← All tools