Batch API charged at regular API prices instead of discounted batch pricing

Gemini API
, ,
1

Hi!

I am experiencing a billing issue with the Gemini Batch API. Despite using the Batch API correctly, I am being charged regular API prices instead of the documented 50% batch discount.

Account Information

  • Account Tier: Tier 2
  • API: Gemini Developer API
  • SDK: google-genai Python SDK (version 1.50.1)
  • Model: gemini-2.5-flash
  • Billing Period: November 15-16, 2025
  • API Key (last 4 chars): …LsM

Issue Details

1. What I Did

I correctly implemented Batch API using the official SDK:

from google import genai

client = genai.Client(api_key=API_KEY)
batch_job = client.batches.create(
    model="gemini-2.5-flash",
    src=requests  # List of InlinedRequest objects
)

2. API Confirmation

Through the API, I can confirm:

  • 181 batch jobs successfully processed
  • All jobs show state: JOB_STATE_SUCCEEDED
  • All jobs used batch API endpoint (client.batches.create())
  • Total: ~221,055 requests processed

3. Billing Shows “BatchGenerate” SKU

My billing report shows the correct SKU indicating Batch API usage:

  • SKU 23BA-BCE4-8C83: BatchGenerate content output token count
  • SKU 6E0D-66C1-4917: BatchGenerate content input token count

4. BUT Prices Are Wrong!

According to the official documentation:

“Batch API is priced at 50% of the standard interactive API cost”

Expected Batch API Pricing:

  • Input tokens: €0.05 per 1M tokens
  • Output tokens: €0.15 per 1M tokens

Actual Charged Pricing:

  • Input tokens: €0.1286 per 1M tokens (2.57x more expensive)
  • Output tokens: €1.073 per 1M tokens (7.15x more expensive!)

Cost Breakdown

Token Usage:

  • Input tokens: 103,485,319 (~103M)
  • Output tokens: 341,078,067 (~341M)
  • Cached tokens: 131,655,646 (~132M)

Expected Cost (with 50% batch discount):

Input:  103M × €0.05/M  = €5.15
Output: 341M × €0.15/M  = €51.15
Cache:  132M × €0.025/M = €3.29
────────────────────────────────
TOTAL:                    €59.59

Actual Billed:

Input:  103M × €0.1286/M = €13.31
Output: 341M × €1.073/M  = €365.92
Cache:  132M × €0.025/M  = €3.38
─────────────────────────────────
TOTAL:                     €382.61

Overcharge: €323.02

Questions for Support

  1. Why is the batch discount not being applied?

    • The SKU clearly shows “BatchGenerate”
    • The API confirms batch job processing
    • But prices are 2-7x higher than documented

  2. Does Tier 2 have different batch pricing?

    • If yes, where is this documented?
    • The public documentation shows 50% discount for all tiers

  3. Is “short input text” SKU priced differently?

    • My billing shows “gemini 2.5 flash short input text”
    • Does this have premium pricing even for batch API?

Requested Resolution

  1. Confirm correct batch API pricing for Tier 2 accounts
  2. Apply retroactive credit/refund for the overcharge (€323.02)
  3. Ensure future batch jobs are billed at correct batch API rates
  4. Provide documentation if Tier 2 has different pricing

Please investigate this urgently as it significantly impacts my project budget. I have paused all further batch processing until this is resolved.

Thank you for your assistance.

Best regards

3

Hi @Denis1 ,

Welcome to the Forum!
For any billing-related issues, please reach out to Google Cloud Billing Support .