We are experiencing a significant billing discrepancy while using the Gemini 2.5 Pro model. According to the official pricing:
Input price: $1.25 per 1M tokens (for prompts ≤ 200k tokens)
Output price (including thinking tokens): $10.00 per 1M tokens (for prompts ≤ 200k tokens)
However, based on our Google Cloud billing data for July 23, the charges appear to be much higher than expected.
Here are the details:
Generate content output token count (Gemini 2.5 Pro short output text): 124,845 tokens → Charged $36.34
Generate content input token count (Gemini 2.5 Pro short input text): 46,007 tokens → Charged $1.67
Generate content input token count (Gemini 2.5 Pro input image): 26,832 tokens → Charged $0.98
Total billed amount: $38.99
According to the official pricing, the calculation should be:
Input text: 46,007 × $1.25 / 1,000,000 = $0.0575
Input image: 26,832 × $1.25 / 1,000,000 = $0.03354
Output text: 124,845 × $10 / 1,000,000 = $1.24845
Expected total cost: ~$1.34
This results in an approximate 30x difference compared to the billed amount.
Could you please help us investigate this issue and clarify the reason for this discrepancy?
Thank you for your support.
Google Cloud report: Custom range, Group by SKU
