Supported Gemini 2.5 Flash access for a separate overseas project of an existing application

Gemini API
1

We operate one mobile application and are preparing a shared overseas environment, starting with Japan. We need to preserve an existing Korean environment and keep the same model assignments while separating overseas credentials, data and project quotas.

Our new overseas evaluation project cannot generate with gemini-2.5-flash, although its model metadata is accessible. The existing service project succeeded with the exact model in an earlier controlled test. We are not asking to change or share that production project’s keys, quotas or configuration.

The latest recorded reproduction was on October 13, 2026 at 11:47:16 UTC:

  • GET https://generativelanguage.googleapis.com/v1beta/models/gemini-2.5-flash: HTTP 200, with generateContent support listed.
  • POST https://generativelanguage.googleapis.com/v1beta/models/gemini-2.5-flash:generateContent: HTTP 404 / NOT_FOUND using the same project-specific credential.
  • Authentication: x-goog-api-key header. API: direct REST, not a mobile SDK or third-party proxy.
  • The test used one generic Japanese confirmation sentence, maxOutputTokens: 8 and thinkingBudget: 0. No customer information or application prompt was sent. We stopped after the error.
  • AI Studio showed paid Tier 1 / Prepay for the evaluation project, and we rechecked that display on October 16. We did not run another generation request on October 16. A billing tier display is not proof of model access.

We have read the September 18 access update:

We also reviewed the similar metadata-200 / generation-404 discussion:

Could the Gemini API team please clarify:

  1. Can an existing developer’s prior Gemini 2.5 usage qualify a newly isolated project for access, or is eligibility strictly tied to the original project or authorization principal?
  2. Is there an official review or exception process to enable the exact gemini-2.5-flash model on a separate project, while leaving the existing service environment unchanged?
  3. If that is not supported, please confirm that explicitly so we can make an informed product decision. We do not want to assume that another key, more prepayment or broader IAM permissions will resolve the restriction.
  4. Which private support channel should we use for the project ID/number and other non-secret diagnostic details? What additional evidence is required?

We understand that a model listing does not establish generation eligibility and that the Gemini Developer API and Vertex model lifecycles are separate. Please route this to the relevant Gemini Developer API access team if possible.

For privacy, this public post does not include project IDs, account emails, billing identifiers, credentials, customer data, proprietary prompts or raw logs. We can provide the affected project’s identifiers through an official private channel if requested. We did not retain a request ID or raw error text and will not invent them.

2

Hello @Nabie ,

To ensure reliable performance for everyone, we are limiting access to the 2.5 models to users who have actively used them in the past. These models are not deprecated and will continue to be served until further notice. For your new projects, please use our latest models: 3.5 Flash-Lite or 3.8 Flash. This helps us maintain sufficient capacity for both ongoing legacy workflows and new applications.