Auto-upgrade
An auto-upgrade is the host moving your deployment onto a newer model version without you changing the model string. Azure documents this for Standard deployment types, provisioned deployments are excluded, and a Gemini latest alias is a different swap.
Last reviewed . Publisher: SunsetDiff. A source line names the day that page was read.
The rows below are copied from this site's model records. A shutdown cell is the date stored for that provider, not a date copied from another host.
Azure OpenAI
For Global Standard, Data Zone Standard, and Standard deployments, Microsoft manages automatic upgrades when a model version is retired. Upgrades are scheduled region by region. The schedule is the model retirement schedule.
versionUpgradeOption has three values. OnceNewDefaultVersionAvailable upgrades when a new default is set. OnceCurrentVersionExpired upgrades only at retirement. NoAutoUpgrade never auto-upgrades, and the deployment stops working at retirement. The working-with-models page says null means OnceCurrentVersionExpired. Automatic updates are documented for Standard deployment types.
Provisioned deployments are not auto-upgraded. You migrate them yourself. The lifecycle page offers an in-place migration, with Azure moving traffic over a 20 to 30 minute window, or a side-by-side migration where you create a second deployment.
The lifecycle page, checked October 22, 2026, uses gpt-4o version 2024-05-13 as the example: on the Standard SKU it retires on 2026-12-09 and the service upgrades it to gpt-5.6-sol where that version is available. The same page says to recheck the retirement schedule, because the date and the replacement can change.
Sources: Foundry Models lifecycle , Working with Azure OpenAI models , Azure retirement schedule , checked October 22, 2026.
Gemini API
A latest alias is hot-swapped on every new release of that variation. Breaking changes get two weeks of email. That is not an Azure deployment upgrade option. The deprecations page also records some Gemini rows whose endpoint stays up and automatically routes requests to another id, with no shutdown date.
Sources: Gemini API models , Gemini API deprecations , checked October 20, 2026.
OpenAI, Anthropic, and Amazon Bedrock
OpenAI moves a short alias by pointing it at a new snapshot. That is a name change on OpenAI's side, not a deployment upgrade setting. Anthropic does not update the weights of an existing model ID. A new model ships under a new ID.
Bedrock says migration will not happen automatically. You change the model id before the EOL date.
Sources: OpenAI changelog , Claude model IDs and versioning , Bedrock model lifecycle , checked October 22, 2026.
Rows from this dataset
The Azure row named in the lifecycle example, when it is in the dataset.
| Model | Provider | Stored status | Shutdown | Replacement |
|---|---|---|---|---|
gpt-4o (2024-05-13) gpt-4o-2024-05-13 | Azure OpenAI | deprecated | 2026-12-09 | gpt-5.6-sol · diff |
| gpt-5.6-sol | Azure OpenAI | active | Not published | Not published |
Gemini API rows whose notes say requests are auto-routed, with no shutdown date.
| Model | Provider | Stored status | Shutdown | Replacement |
|---|---|---|---|---|
| gemini-3.5-flash Release date on the deprecations page: 2026-05-19. Deprecated, requests auto-routed to gemini-3.6-flash, no shutdown date. | Google Gemini API | deprecated | Not published | gemini-3.6-flash · diff |
| gemini-3.7-flash Release date on the deprecations page: 2026-08-13. Deprecated, requests auto-routed to gemini-3.8-flash, no shutdown date. | Google Gemini API | deprecated | Not published | gemini-3.8-flash · diff |