Share via

Data EU residency of Cohere Rerank 4 serverless

Lukáš Krch 20 Reputation points
2026-05-14T15:37:08.9066667+00:00

Hi,

I am planning to use Cohere Rerank 4 via Serverless API (MaaS) in Sweden Central. Regarding EU data residency, please confirm:

Inference Boundary: Does data processing for inference stay strictly within Sweden Central (or the EU)?

Global Routing: Is there any "Global" load balancing that could cause data to be processed outside the EU?

Data at Rest: Can you confirm all logs or metadata remain within the regional boundary?

I need to ensure that no data leaves the EU during the entire request-response lifecycle.

Thank you.

Microsoft Security | Microsoft Purview
0 comments

Answer accepted by question author

  1. Sina Salam 28,931 Reputation points Volunteer Moderator
    2026-05-17T06:18:01.8666667+00:00

    Hello Lukáš Krch,

    Welcome to the Microsoft Q&A and thank you for posting your questions here.

    I understand that you are planning to ensure strict EU data residency when using Cohere Rerank v4.0 via serverless API in Sweden Central.

    The key issue is that data residency is not determined by the region (Sweden Central) alone, but by the deployment type (SKU) used in Azure AI Foundry.

    Therefore, to meet the requirement that no data leaves the EU during the entire lifecycle:

    • Do NOT use GlobalStandard or any Global deployment
    • Use either:
      • EU DataZone deployment (EU-wide processing allowed)
      • *Regional Standard deployment in an EU region (preferred if available)
    • If Cohere Rerank v4.0 only supports Global deployment in your environment, then it cannot meet the EU-only requirement, and an alternative model/deployment must be used

    All the resource links associated are for more reading and steps.

    I hope this is helpful! Do not hesitate to let me know if you have any other questions, steps or clarifications.

    Please don't forget to close up the thread here by upvoting and accept it as an answer if it is helpful.

    Was this answer helpful?

    3 people found this answer helpful.

1 additional answer

Sort by: Most helpful
Most helpful Newest Oldest
  1. Anshika Varshney 11,060 Reputation points Microsoft External Staff Moderator
    2026-05-18T13:14:09.1566667+00:00

    Hello Lukáš Krch,

    The important thing to understand here is that data residency for Cohere rerank (serverless API) is not decided only by the region you select, but mainly by the deployment type used in Azure AI Foundry.

    How data residency works in this case

    There are different deployment types, and each one behaves differently:

    1. Global deployments

    • Requests can be processed in any Azure region
    • This means data may go outside the EU
    • Not suitable if you need strict EU-only processing

    2. EU DataZone deployments

    • Data processing stays within the EU region
    • But it is not limited to a single region like Sweden Central
    • Processing may move between EU regions

    3. Regional (standard) deployments

    • Data processing happens in the selected region only
    • This is the best option if you need strict EU residency
    • Example, if deployed in Sweden Central, processing stays there

    Important note about serverless APIs

    Serverless models (pay as you go) are also tied to these deployment types. So even if you select a region like Sweden Central:

    • If it uses global routing, data may still leave the EU
    • If it is regional deployment, it will stay in that region

    About data at rest

    • Logs, metadata, and stored data remain in the selected Azure geography
    • But this does not control where inference processing happens

    What you should do

    If your requirement is strict EU compliance, check the following:

    • Make sure you are using regional deployment type
    • Avoid global deployments
    • Confirm your model endpoint is created under a region that supports the model (for example Sweden Central is supported for Cohere models) [docs.cohere.com]

    Reference documentation

    You can review these for more details:

    Summary

    • Region selection alone is not enough
    • Deployment type decides where inference runs
    • For strict EU-only processing, use regional deployment

    I Hope this helps. Do let me know if you have any further queries.

    If this answers your query, please do click Accept Answer and Yes for was this answer helpful.

    Thankyou!

    Was this answer helpful?

    1 person found this answer helpful.
    0 comments

Your answer

Answers can be marked as 'Accepted' by the question author and 'Recommended' by moderators, which helps users know the answer solved the author's problem.