Generative AI Regions

Learn where OCI Generative AI hosts models and how to access the service from CPU-only regions.

Oracle hosts its OCI services in regions and availability domains. A region is a localized geographic area, and an availability domain is one or more data centers in that region.

Important

See Generative AI Models by Region for model availability in the regions that serve inference requests. For regions with CPU-only access to on-demand inference, see Commercial Regions (OC1) with CPU-Only Access.

Commercial Regions (OC1) with Model Hosting

The following table lists the OC1 regions where Generative AI hosts models. Available models and serving modes vary by region. To use an on-demand model that isn't available locally, create a routing profile for that model and select regions where it is available. Use the profile from your current region in the Playground or through a supported inference API.

Region Name Location Region Identifier Region Key
Brazil East (Sao Paulo) Sao Paulo sa-saopaulo-1 GRU
Germany Central (Frankfurt) Frankfurt eu-frankfurt-1 FRA
India South (Hyderabad) Hyderabad ap-hyderabad-1 HYD
Japan Central (Osaka) Osaka ap-osaka-1 KIX
Saudi Arabia Central (Riyadh) Riyadh me-riyadh-1 RUH
UAE Central (Abu Dhabi) Abu Dhabi me-abudhabi-1 AUH
UAE East (Dubai) Dubai me-dubai-1 DXB
UK South (London) London uk-london-1 LHR
US East (Ashburn) Ashburn us-ashburn-1 IAD
US Midwest (Chicago) Chicago us-chicago-1 ORD
US West (Phoenix) Phoenix us-phoenix-1 PHX

Learn About Regions and Availability Domains.

Commercial Regions (OC1) with CPU-Only Access

For the OC1 regions that support CPU-only access to on-demand inference in Generative AI, see Commercial Regions (OC1) with CPU-Only Access. This list includes UAE Central (Abu Dhabi), which also hosts models for dedicated AI clusters. For all OCI commercial regions, see the OCI regions table and refer to entries with the realm key OC1.

In a CPU-only region, create a routing profile for the on-demand model you want to use and select target regions where it is available. Select the profile in your region's Playground or reference it in an API request to run the model in a selected target region. The Chat API, OpenAI-compatible Chat Completions API, and Responses API are supported when the selected model supports the API. Regional API endpoints support IPv6.

For supported features, limitations, and routing instructions, see Generative AI in CPU-Only Regions.

Government Region (OC4)

Region Name Location Region Identifier Region Key
UK Gov South (London) London uk-gov-london-1 LTN

Learn about Oracle UK Sovereign Cloud.

Sovereign Region (OC19)

Region Name Location Region Identifier Region Key
EU Sovereign Central (Frankfurt) Frankfurt eu-frankfurt-2 STR

Learn about Oracle EU Sovereign Cloud.

Services that Call into the Generative AI Service

Some OCI services call the Generative AI service. The following table lists the calling and destination regions for those service integrations.

There are OCI services which document (in the “Oracle PaaS and IaaS Universal Credits Service Descriptions”) that they make calls into the OCI Generative AI Service. For a given call into the OCI Generative AI Service, if the Calling Region and Destination Region are not the same, then a cross-region call will be made.

Area

Calling Regions:

OCI regions from which (cross-region) calls can be made to the OCI Generative AI Service

Destination Region:

OCI region offering OCI Generative AI Service

Brazil GRU, VCP GRU
EU AMS, MRS, LIN, ARN, CDG, MAD, ZRH, FRA FRA
India BOM, HYD HYD
Japan KIX, NRT KIX
Saudi Arabia RUH RUH
UAE DXB DXB
UK CWL, LHR LHR
USA PHX, IAD, SJC, ORD ORD