Azure OpenAI retirements from January 12, 2027 to February 9, 2027
The dataset has 5 Azure OpenAI ids in this window. 2027-01-12: gpt-chat-latest-2026-08-06 (Preview). 2027-01-14: gpt-4o-2024-05-13 (Deprecated). 2027-01-20: gpt-realtime-2-2026-05-06 (Preview). 2027-02-09: gpt-5-2025-08-07 (GA), gpt-5-nano-2025-08-07 (GA). The stored replacement is filled in for gpt-realtime-2-2026-05-06. The replacements list is empty for gpt-chat-latest-2026-08-06, gpt-4o-2024-05-13, gpt-5-2025-08-07, gpt-5-nano-2025-08-07, so those cells say Not published. The lifecycle page says a Retired model is removed from service and all inference requests return 410 Gone. Standard deployments can auto-upgrade. A deployment set to NoAutoUpgrade stops working. Provisioned deployments are not auto-upgraded.
Last reviewed . Publisher: SunsetDiff.
Steps
Read the date table. The table is these 5 Azure OpenAI ids. The lifecycle cell is the schedule's Lifecycle column, copied from the source excerpt stored on the row. Shutdown date and replacement are the fields stored on the row. Not published means the replacements list is empty. Open the model page before you change a deployment.
List deployments on the resource. Call Deployments List, or Get-AzCognitiveServicesAccountDeployment, for each Azure OpenAI resource. Read properties.model.name and properties.model.version. A deployment name can differ from the model name.
Read versionUpgradeOption. OnceCurrentVersionExpired upgrades the deployment when the retirement date is reached. null means the same thing. NoAutoUpgrade never upgrades, and that deployment stops working. Provisioned deployments are not auto-upgraded.
Use the stored replacement, or shortlist one. Where the table names a replacement, that id is the one stored on the row. Where the cell says Not published, the replacements list is empty. The migration page asks: "Is there a named suggested replacement, or do you need to shortlist candidates yourself?" Do not wait for a target the row does not store. Deploy the model you picked and send one real request before the shutdown date on that row.
Models and dates
These 5 ids are the Azure OpenAI rows in this guide. The model id links to its SunsetDiff model page. Lifecycle is the schedule's Lifecycle column, copied from the source excerpt stored on the row. Shutdown date and replacement are the fields stored on the row. Not published means the replacements list is empty. These rows do not store a deprecation_date. 2027-01-12: gpt-chat-latest-2026-08-06 (Preview). 2027-01-14: gpt-4o-2024-05-13 (Deprecated). 2027-01-20: gpt-realtime-2-2026-05-06 (Preview). 2027-02-09: gpt-5-2025-08-07 (GA), gpt-5-nano-2025-08-07 (GA).
The schedule says: "The Replacement column identifies Microsoft's recommended replacement and, where supported, the target for automatic upgrades at retirement." This guide does not add a replacement the stored list does not have.
5 Azure OpenAI model ids retiring from 2027-01-12 through 2027-02-09.
The schedule says: "The Replacement column identifies Microsoft's recommended replacement and, where supported, the target for automatic upgrades at retirement." It also says: "You don't need to wait for a replacement to be listed before evaluating or manually migrating to another compatible model." The lifecycle page says: "The replacement listed in the Model retirement schedule is Microsoft's recommended migration target and, where supported, the model used for automatic upgrades at retirement. It doesn't limit your manual migration choices."
The replacements list is filled in for gpt-realtime-2-2026-05-06. The recommended replacement is the official replacement stored on each row. gpt-realtime-2-2026-05-06: gpt-realtime-2.1.
The replacements list is empty for gpt-chat-latest-2026-08-06, gpt-4o-2024-05-13, gpt-5-2025-08-07, gpt-5-nano-2025-08-07. Empty means the row does not store an official replacement. This guide does not add one.
gpt-4o-2024-05-13: the replacements list is empty, so the table says Not published. The source excerpt on that row ends with gpt-5.1 (2025-11-13). The lifecycle page says: "All active customer-facing base-model inference offers for gpt-4o version 2024-05-13 in Azure public cloud retire on January 14, 2027, in all applicable regions." Then: "The replacement model is gpt-5.1 version 2025-11-13." SunsetDiff does not store that excerpt cell as an official replacement id. The dataset does store gpt-5.1-2025-11-13 with shutdown date 2027-07-13 and replacement Not published.
The stored replacement id gpt-realtime-2.1 is gpt-realtime-2.1. Its stored shutdown date is Not published.
gpt-realtime-2.1-2026-07-07 is a different id. It stores shutdown date 2027-06-25 and replacement Not published. The retiring row does not name that dated id as its replacement.
The lifecycle table says Retired means: "Removed from service. All inference requests return 410 Gone." New deployments are not allowed, and existing deployments do not work. The same page says the API value lifecycleStatus Deprecated means the model is retired and inference returns 410 Gone. The docs stage Deprecated is lifecycleStatus Deprecating, and inference still works. The sentences and the status map are on Azure OpenAI retired models return 410 Gone.
gpt-4o-2024-05-13 is Deprecated on the schedule now, so inference still works until the shutdown date stored on that row.
gpt-chat-latest-2026-08-06 is Preview on the schedule, and the replacements list is empty. The lifecycle page says: "Preview models aren't recommended for production workloads." It also says all preview deployments are either upgraded or terminated. And: "If no replacement exists, customers get 30 days notice before the model retires and inference returns 410 Gone." The schedule does not store a replacement, so this page does not pick the upgrade path or the no-replacement path for that preview row.
gpt-realtime-2-2026-05-06 is Preview on the schedule, and the replacements list is filled in. The lifecycle page says: "Preview models aren't recommended for production workloads." The replacement column is the target where an automatic upgrade is supported.
gpt-5-2025-08-07, gpt-5-nano-2025-08-07 are GA on the schedule. The lifecycle page says: "For Global Standard, Data Zone Standard, and Standard deployment types, Microsoft manages automatic upgrades when a model version is retired:" The migration page says: "Standard, Global Standard, and Data Zone Standard (pay-as-you-go) deployments are auto-upgraded on a rolling, region-by-region schedule." The replacement column is the target where an automatic upgrade is supported. These GA rows have an empty replacements list, so this page does not name the model an auto-upgrade would select.
The lifecycle page says: "Provisioned deployments are NOT auto-upgraded. Provisioned customers must manually migrate to the replacement model." The migration page says: "Provisioned (PTU) deployments are not auto-upgraded." And: "Batch deployments follow the side-by-side path: deploy the new model, resubmit jobs, retire the old deployment."
Working with models says: "Automatic model updates are only supported for Standard deployment types." OnceCurrentVersionExpired means: "Once the retirement date is reached the model deployment automatically upgrades to the current default version." NoAutoUpgrade means: "The model deployment never automatically upgrades. Once the retirement date is reached the model deployment stops working." And: "null is equivalent to OnceCurrentVersionExpired."
The same page says: "When you select a specific model version for a deployment, this version remains selected until you either choose to manually update yourself, or once you reach the retirement date for the model." Then: "When the retirement date is reached the model will automatically upgrade to the default version at the time of retirement." A deployment you pinned to one of these versions can still move on the retirement date unless versionUpgradeOption is NoAutoUpgrade. NoAutoUpgrade does not move, and the deployment stops. That move still needs a default version. A row with an empty replacements list does not name one.
The migration page says: "You also need an inventory of the model deployments your workload uses, including their deployment types (Standard, Provisioned, or Batch)." Working with models says: "To query the current model deployment settings including the deployment upgrade configuration for a given resource use Deployments List." If the value is null, you won't see a versionUpgradeOption property. The model versions page says: "Go to your deployment in the Foundry portal, then open the Details tab to see the current model version." And: "Query the deployments endpoint for your resource. The response includes the model version for each deployment."
The example response puts the model name and version on properties.model. The deployment's own name is the name field next to that object. Match properties.model.name and properties.model.version to the ids in the table.
Also search the repository. A hit is a candidate, not proof a deployment still uses that version. How to find every model ID in your codebase covers SDK argument names. The scan page builds a search from dataset ids that have a shutdown date.
Azure PowerShell, list deployments for one account
GET https://management.azure.com/subscriptions/{subscriptionId}/resourceGroups/{resourceGroupName}/providers/Microsoft.CognitiveServices/accounts/{accountName}/deployments?api-version=2023-05-01
OpenAI's API is a different provider. The dataset has 9 OpenAI ids with shutdown date 2027-01-14. They are on GPT-4.1 and GPT-4o shutdown on January 14, 2027. That date is not an Azure deployment date.
SunsetDiff publishes a shutdown batch page when at least three model records share one provider and one exact shutdown date. A named date with fewer rows can still have a page.
The recommended replacement is the official replacement stored on each row. gpt-chat-latest-2026-08-06: Not published. gpt-4o-2024-05-13: Not published. gpt-realtime-2-2026-05-06: gpt-realtime-2.1. gpt-5-2025-08-07: Not published. gpt-5-nano-2025-08-07: Not published. The schedule says: "The Replacement column identifies Microsoft's recommended replacement and, where supported, the target for automatic upgrades at retirement." The replacements list is empty on gpt-chat-latest-2026-08-06, gpt-4o-2024-05-13, gpt-5-2025-08-07, gpt-5-nano-2025-08-07.
Why does gpt-4o 2024-05-13 say Not published?
gpt-4o-2024-05-13 stores no official replacement, so the table says Not published. The source excerpt on that row ends with gpt-5.1 (2025-11-13). The lifecycle page says: "The replacement model is gpt-5.1 version 2025-11-13." SunsetDiff does not store that excerpt cell as an official replacement id. The dataset stores gpt-5.1-2025-11-13 at 2027-07-13 with replacement Not published.
What does Azure return after retirement?
The lifecycle page says a Retired model is removed from service and: "Removed from service. All inference requests return 410 Gone." See Azure OpenAI retired models return 410 Gone.
Will every deployment upgrade itself?
No. The lifecycle page says: "For Global Standard, Data Zone Standard, and Standard deployment types, Microsoft manages automatic upgrades when a model version is retired:" And: "Provisioned deployments are NOT auto-upgraded. Provisioned customers must manually migrate to the replacement model." Working with models says: "Automatic model updates are only supported for Standard deployment types." NoAutoUpgrade means: "The model deployment never automatically upgrades. Once the retirement date is reached the model deployment stops working." null is equivalent to OnceCurrentVersionExpired. The rows with an empty replacements list do not name an upgrade target: gpt-chat-latest-2026-08-06, gpt-4o-2024-05-13, gpt-5-2025-08-07, gpt-5-nano-2025-08-07.
How do I find deployments that still use these versions?
Working with models says: "To query the current model deployment settings including the deployment upgrade configuration for a given resource use Deployments List." Read properties.model.name and properties.model.version. The model versions page says: "Go to your deployment in the Foundry portal, then open the Details tab to see the current model version." The migration page says: "You also need an inventory of the model deployments your workload uses, including their deployment types (Standard, Provisioned, or Batch)."
Is there a shutdown page for each of these dates?
A shutdown batch page is published when at least three model records share one provider and one exact shutdown date, or when the date is explicitly included. January 12, 2027 has 1 Azure OpenAI id with that exact shutdown date, so the batch page is Azure OpenAI shutdowns on January 12, 2027. January 14, 2027 has 1 Azure OpenAI id with that exact shutdown date, so the batch page is Azure OpenAI shutdowns on January 14, 2027. January 20, 2027 has 1 Azure OpenAI id with that exact shutdown date, so the batch page is Azure OpenAI shutdowns on January 20, 2027. February 9, 2027 has 2 Azure OpenAI ids with that exact shutdown date, so the batch page is Azure OpenAI shutdowns on February 9, 2027.
Is this the same shutdown as OpenAI's January 2027 guides?