Unexpectedly High Minimum Charges for Gemini 1.5 Flash API Usage

Hello Google Cloud Community,

I’m reaching out for some clarification regarding the billing for Gemini 1.5 Flash API usage within Vertex AI Generative AI. I’m currently evaluating Gemini for a project, and while I’m in a trial period (so not being charged yet), the projected costs in my billing report seem disproportionately high for very low usage. This is making it difficult to accurately estimate future costs.

I’ve attached a screenshot of my billing report for July 2025 (july.png) to illustrate the issue.

Here’s a summary of my usage and the reported costs for July 2025:

  • Input Tokens (SKU: GenerateContent input token count for Gemini 1.5 Flash when input is up to 128k tokens, SKU ID: D8AE-AF3C-415B):

    • Reported Usage: 5,893 input tokens.

    • Reported “Cost per use”: $2 USD.

    • Expected cost based on official pricing (USD 0.075 per 1 million tokens): Approximately $0.00044 USD.

  • Output Tokens (SKU: GenerateContent output token count for Gemini 1.5 Flash when input is up to 128k tokens, SKU ID: 7DF3-1F04-931A):

    • Reported Usage: 2,372 output tokens.

    • Reported “Cost per use”: $3 USD.

    • Expected cost based on official pricing (USD 0.30 per 1 million tokens): Approximately $0.00071 USD.

As you can see, the “Cost per use” shown in the billing report is thousands of times higher than the published per-token pricing for Gemini 1.5 Flash. My total usage for July was only 5,893 input tokens and 2,372 output tokens, which should cost less than a cent combined. However, the billing report shows a total of $5 USD before trial savings.

My questions to the community are:

  1. Has anyone else experienced similar billing patterns for very low usage of Gemini 1.5 Flash (or other Vertex AI Generative AI models)?

  2. Is there a known minimum charge per SKU or per operation that is applied, even for minimal token consumption, which might not be explicitly detailed in the main pricing documentation?

  3. If so, where can I find clear documentation about these minimum charges or fixed fees?

  4. Will these minimum charges apply once my trial period ends?

Understanding this cost structure is crucial for my project planning. Any insights or experiences you can share would be greatly appreciated.

Thank you in advance for your help!

Best regards,
David Suarez.

This could be similar billing problem with the recent incident. Probably while deploying a model, they somehow broke the billing pipeline imo.

I’m not sure though, since yours is 1.5-flash.

I got charged like $200+ with this problem, somehow i guess they will pay me back the amount taken.

Continuing the discussion from [User Archive] Data export complete: