Gemini 3 - 403 PERMISSION_DENIED - The caller does not have permission

Gemini API
, ,
1

I’m in the middle of doing some changes on one of our AI project that uses gemini api when I suddenly get PERMISSION_DENIED on all Gemini 3 models. Gemini 2 still works fine but what’s that point paying when we’re restricted to using the old models?

{'code': 403, 'message': 'The caller does not have permission', 'status': 'PERMISSION_DENIED'}

I have tried multiple api keys both free and billed and they are all unable to use Gemini 3. We barely reach our tier limit, each of our request use less than 2000 tokens and there are very few request per minutes.

Our prod servers are getting the same response and are falling back to gemini 2 models. If gemini 2 starts throwing the same error, this feature of our app basically becomes unusable.

2

Same, hopefully we get a message soon!

3

Same issue here. Though it was me…

4

Same. We received 3 emails in succession about 8 hours ago upgrading our “tier” in AI studio. We do not use AI studio, API keys made about 11 months ago.

All gemini reliant services are down for us as a result.

5

Oddly enough, Gemini 3 Flash Preview seems to work, unlike other 3.x models.

6

gemini-3.5-flash and gemini-3.7-flash also works for me now. gemini-3.6-flash still returns 403. They are most probably working on it right now. But sucks I have to baby sit for now till I finish automating things.

7

Same issue here. 403 errors have exploded today

8

Hello Everyone,

if you are still experiencing this issue. Could you please provide a bit more information via DM so we can investigate further?

  1. The GCP Project ID (or Project Number)
  2. The target API endpoint that you are trying to hit

To send a DM (Direct Message), click on my profile picture or name, and then select the Message button.

9

Same as @Elie_Rotenberg mentioned.
gemini-2.5-flash and gemini-3-flash-preview are working fine, unlike any other 3.x models.

10

From Yesterday, I have experienced the same issue. gemini-3-flash-preview is still working, but other 3.x models(gemini 3.1 pro preview and gemini 3.5 flash lite) are not.

In my situation, with file attachment via external file link, this erros happen.

403 PERMISSION_DENIED only on generateContent with file_data/file_uri — uploads succeed, text/inline succeed (prepay account, key restricted to Gemini API)

11

Hi!

I am experiencing the exact same issue here. In AI Studio the models gemini-3.8-flash, gemini-3.5, and gemini-3.1-flash-lite work fine under my project, but gemini-3.6-flash consistently throws a 403: The caller does not have permission error. I am using plain text prompts only (no files, no grounding, no code execution sandbox active). Tried starting new chat, new project, incognito tab, etc and nothing works.

This is heavily reducing my coding workflow since I rely on 3.6 for deep reasoning/coding capabilities. 3.5 and 3.1 flash lite don´t work.

Could an engineer please check if there is an isolated provisioning or routing bug affecting the 3.6 Flash endpoint for free-tier users? Thanks!

Error Gemini 3.6 flash
Error Gemini 3.6 flash503×324 21.5 KB

12

Hello,

In Google AI Studio, gemini-3.6-flash requires either a linked paid API key (from a project with Cloud Billing or prepayment credits enabled) or a Google AI Pro / Google AI Ultra subscription. Free-tier keys are only supported in the AI Studio Playground on models that include free Playground access, such as gemini-3.8-flash, gemini-3.5-flash-lite, and gemini-3.1-flash-lite.

To continue without 403 PERMISSION_DENIED errors, you can do any of the following:

  • Use gemini-3.8-flash, gemini-3.5-flash-lite, or gemini-3.1-flash-lite free of charge in Google AI Studio.
  • Link a paid API key (from a project with Cloud Billing or prepayment credits enabled) in Google AI Studio (Billing documentation).
  • Call gemini-3.6-flash directly through the Gemini API (https://generativelanguage.googleapis.com) using your API key, where Free Tier rate limits remain available (Gemini 3.6 Flash documentation).