Exa Search, now in the EU

Exa Search, now in the EU
The Exa Team
The Exa Team
Nov 13, 2026

Starting today, customers can run Exa search inside the European Union, on our main web index, with the same models our US customers use. Queries, results, and page contents are stored and processed in AWS Stockholm (eu-north-1).

Companies and AI products with customers in Europe often can’t send queries to a US service, either due to GDPR data residency requirements, or to internal requirements from their procurement, security, and regulatory teams. A significant portion of Exa’s API traffic already comes from Europe, and many of our customers have requested support for EU data residency.

Serving Exa from the EU

To serve Exa from the EU, we built a true second home from the ground up. Every system that touches your data runs regionally:

  • We replicated our main web index into the EU. Petabytes of data, maintained in real time across two continents.
  • API routers, vector databases, inference, rerankers and highlights services all run on our own GPUs in Stockholm.
  • Contents are retrieved from the EU and fetched via EU endpoints.

Same quality as US

As of today the EU region serves the Search and Contents API endpoints, with the auto and lite search types. We compared results across a series of standard benchmarks to ensure that search results are at the same quality in the US and EU.

Exa Lite: US vs EU
US liteEU lite
BenchmarkScoreΔ EU − US
deep-search-qa (F1)
0.867
0.867
0.000
0.455
0.479
+0.024
live-browsecomp
0.546
0.560
+0.014
finsearchcomp
0.665
0.685
+0.020
Exa Auto: US vs EU
US autoEU auto
BenchmarkScoreΔ EU − US
deep-search-qa (F1)
0.876
0.880
+0.004
0.497
0.469
−0.028
live-browsecomp
0.556
0.560
+0.004
finsearchcomp
0.655
0.655
0.000

For all benchmarks there were no more than three percentage points of difference between scores for the US and EU endpoint. When searching from the EU, using the nearer EU endpoint also saved more than 100ms of latency compared to the US.

Where your data lives

Customer content stays in the EU. Requests that can’t be served in the EU are refused, with no silent fallback to the US. Your queries, the results we return, and the page contents all stay in the EU. Every EU team runs with zero data retention by default, so request content isn't kept at all. TLS connections to api.eu.exa.ai are terminated only in EU data centers.

How we enforce it:

  • Region-locked accounts. An EU team's API key works only on the EU endpoint and is refused everywhere else. If a client is misconfigured, the request fails instead of quietly failing over to the US. An EU team with an EU key can’t make a mistake and accidentally use the US endpoint.
  • No silent fallback. If a feature isn't available in the EU yet, we refuse the request before it's metered. We never serve it from another region.
  • Each response carries an x-exa-processing-region: eu header, so you can audit where every request was processed.
  • Every residency rule is baked into our codebase and checked in code review. Monitoring alerts us if EU data ever shows up where it shouldn't.

Account and identity metadata and aggregated service metrics may be kept outside the EU.

Search for the world

This work doesn't stop in Europe, we’re building for the world. We'll continue to serve Exa in more regions. More search types are coming soon to the EU, and Agents, Websets, and Monitors will follow.

Getting started

Point your client at api.eu.exa.ai and talk to us about an EU team.

from exa_py import Exa exa = Exa(api_key="YOUR_EU_KEY", base_url="https://api.eu.exa.ai") results = exa.search("EU AI Act obligations for model providers", type="auto")

EU access comes with a new EU team, set up by your Exa rep. A team's region is set when the team is created and doesn't change, which is part of what keeps your data where you put it.

Talk to sales →