Gemini-2.5-flash visible via models.get and advertises generateContent, but generateContent returns 404 / MODEL_OR_RESOURCE

Gemini API
1

Hello Gemini API team,

I am requesting review of a Gemini 2.5 Flash access issue affecting an existing production application.

Project ID: safe-homes-with-toni
Model: gemini-2.5-flash
Credential type: Google AI Studio Auth key, bound to a service account
API restriction: Gemini API only
Application restriction: None

The server-side credential is correctly configured and reaches the Gemini API.

A read-only Models API request using the same production credential returns:

MODELS_GET_HTTP=200
MODEL_GET_FOUND=YES
MODEL_NAME_MATCH=YES
GENERATE_CONTENT_SUPPORTED=YES

Therefore, the API exposes models/gemini-2.5-flash to this credential and explicitly advertises generateContent as supported.

However, a production generateContent request using the same credential and the exact model gemini-2.5-flash returns HTTP 404, classified by our sanitized server diagnostics as:

MODEL_OR_RESOURCE

We performed only one generation attempt and did not retry it.

We have independently verified:

  • the API key is associated with the correct Google Cloud project;
  • it is an Auth/service-account-bound key;
  • the Gemini API is the selected API restriction;
  • the bound service account exists;
  • the key is not shown as blocked or disabled;
  • the runtime secret is correctly available to the production server;
  • models.get succeeds with HTTP 200;
  • the requested model name is exactly gemini-2.5-flash;
  • supportedGenerationMethods includes generateContent.

No API keys, bearer tokens, prompts, client data, or other credentials are included in this report.

Google documentation currently lists gemini-2.5-flash as a stable model with no announced shutdown date. We are also aware of the recent statement that access to Gemini 2.5 models may be limited based on prior usage.

Could you please confirm whether safe-homes-with-toni or its new Auth-key/service-account identity is currently restricted from Gemini 2.5 Flash generation despite the Models API advertising generateContent support?

If this is an entitlement/provisioning restriction, please advise whether access can be restored for this existing application. We specifically need to remain on gemini-2.5-flash and do not want to substitute another model.

Thank you.

2

Hello @Brandon_Matthews ,

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 of equivalent quality: 3.1 Flash-Lite or 3.5 Flash. This helps us maintain sufficient capacity for both ongoing legacy workflows and new applications.