Gemini 2.5 Pro removed from Gemini Enterprise in Canada and Japan regions
From 20 October 2026, Gemini 2.5 Pro is no longer available in the Gemini Enterprise app for Canada and Japan in-country region users.
From 20 October 2026, users of the Gemini Enterprise web and mobile app in the Canada and Japan in-country regions can no longer select Gemini 2.5 Pro. The model has been retired from those regions, and anyone who was relying on it needs to switch to an alternative.
If you are based elsewhere, nothing changes for you. Gemini 2.5 Pro remains available in the global, us, and eu regions, and administrators in those regions can still toggle it on or off in the web app as before.
What exactly changed, and for whom?
Google retired the gemini-2.5-pro model on 20 October 2026, and its removal from the Gemini Enterprise app in Canada and Japan in-country regions is part of that lifecycle end. The in-country designation refers to users whose Gemini Enterprise tenancy is configured to keep data within Canada or Japan specifically, rather than routing through a broader multi-region.
This change was announced in the Gemini Enterprise release notes on 1 October 2026, giving organisations 19 days to prepare. If you missed that window, the model is now simply gone from the model selector.
It is worth noting that this retirement applies specifically to the Gemini Enterprise app. If you use Gemini 2.5 Pro through the Gemini Developer API directly, Google has not announced a shutdown date on that side.
What should you move to?
Google’s own migration guidance points to a few options, so the picture is slightly more nuanced than “switch to Gemini 4 Argon and you are done.”
Gemini 4 Argon is the flagship successor and became generally available on 13 October 2026. It is a strong replacement for complex, reasoning-heavy work, and Google positions it for demanding tasks across coding, finance, legal, cybersecurity, and scientific domains. It carries the same 1,048,576-token context window as Gemini 2.5 Pro. However, there is a regional catch: Gemini 4 Argon is supported only in the global, us, and eu multi-regions. If your tenancy is locked to a Canada or Japan in-country region, you will need to check whether your routing configuration permits access to those multi-regions before treating Gemini 4 Argon as a straightforward swap.
Gemini 3.8 Flash and Gemini 3.5 Flash are also listed by Google as recommended replacements for Gemini 2.5 Pro users. These are faster and cheaper models rather than straight capability equivalents, but if your workload does not require deep reasoning, they may serve you well and at lower cost. Gemini 3.8 Flash carries a reduced price until 31 December 2026, making it worth a look if budget is a factor.
The honest summary: if you need a like-for-like replacement for complex reasoning tasks, Gemini 4 Argon is the intended successor. If you need something lighter and cheaper, the Flash models are a reasonable step down.
What does this mean for you?
If you are an end user in the Canada or Japan in-country region, you will find Gemini 2.5 Pro simply absent from the model list. You do not need to do anything to “uninstall” it, but you do need to consciously pick a replacement model for any workflows that were using it. Do not assume your prompts are silently falling back to another model without errors.
If you are an IT administrator, the action item is to review which users and workflows in your organisation had Gemini 2.5 Pro selected, and ensure they have a supported model configured. Gemini 4 Argon was released with its default toggle set to off, switching to on by default from 27 October 2026, so you may need to enable it for your users if you have not already. Check the Gemini Enterprise release notes for the toggle details and the Gemini 4 Argon model page for regional availability before committing to it as your standard replacement.
If you are a developer calling gemini-2.5-pro through Google Cloud (the Gemini Enterprise Agent Platform, formerly Vertex AI), the model is retired as of 20 October 2026 and calls to it will fail. Update your model ID to gemini-4-argon or one of the Flash variants, and test your outputs, because these are not identical models and you may see differences in response style or reasoning depth.
A quick note on Gemini Enterprise Agent Platform naming
You may see references to Vertex AI in older documentation. Google has rebranded Vertex AI’s services under the Gemini Enterprise Agent Platform umbrella, so if you are hunting for up-to-date documentation and cannot find what you need in the old Vertex AI docs, the Gemini Enterprise Agent Platform documentation is now the right place to look.
The broader model retirement and migration guidance, including the full list of recommended replacements for Gemini 2.5 Pro, lives on the Gemini 2.5 Pro model page while it remains accessible.