Unexpected 88M Output Tokens Reported for Gemini 2.5 Flash Lite

Hello Google AI Studio Team,

I’m seeing what appears to be an incorrect token usage report in the AI Studio dashboard.

Details:

  • Model: Gemini 2.5 Flash Lite
  • Date: July 23
  • Reported Output Tokens: ~88 million
  • Reported Input Tokens: ~25 million

My application only generates short responses and has relatively low traffic. Based on my application logs and usage patterns, generating approximately 88 million output tokens in a single day is not possible.

I’ve attached a screenshot of the usage dashboard for reference.

Could someone please help clarify:

  1. Is this a known issue with AI Studio usage reporting?
  2. Could this be a token accounting/reporting bug?
  3. Is there any way to view request-level token usage to verify where these output tokens originated?

Has anyone else experienced similar behavior?

Thank you for your help.