Gemini API prepay billing: switch by November 19

Does the November 19 deadline apply to my Gemini API account?

The date comes from a Google AI Studio migration email reviewed by Ganivra. Google’s public guide describes a phased rollout. Check the deadline in your own notice and the billing plan shown for your project. If you already switched and funded the correct account, the email says your migration is complete. Google’s Gemini API billing guide.

How do I switch from postpay to prepay in Google AI Studio?

1. Find the project that owns your production API key. Open Projects in AI Studio. Check its billing plan and linked billing account. If a teammate manages billing, give them the project ID so they can identify the right account.
2. Open the prepay setup. Follow the billing action shown for that project. A project marked Set up Prepay already has a linked billing account. Review the selected account before confirming the change.
3. Complete the payment in the same session. Load at least $5 USD, or the local equivalent shown at checkout. Google warns that leaving setup after confirming the switch but before funding can interrupt existing projects.
4. Verify the balance and make one small request. Check that Billing shows Prepay and a positive available balance. Send a request through your application using the key for the affected project.

Open Google AI Studio Billing · Google’s setup instructions

Is the $5 a fee? How long will it last?

The $5 buys credit toward future Gemini API usage. How long it lasts depends on the combined usage of the projects sharing that balance. For example, if production uses $1.50 a day and staging uses $0.50, a $5 balance covers about two and a half days at that pace.

A top-up also changes how you read your costs. Buying $50 in credits today records a payment. If your applications use $8 of those credits today, $8 is the usage cost to attribute to today’s customers. For the rates behind that calculation, see Gemini API pricing by model.

Should I turn on Gemini API auto-reload?

Google recommends auto-reload in the notice to help keep production requests running. In Billing, open the available credits card and configure auto-reload. Pick a balance threshold and reload amount based on the busiest period you expect across the linked projects. Check that auto-reload shows as enabled. Google’s auto-reload documentation.

For example, if a batch job can use $8 overnight, a $5 opening balance alone cannot cover the run. Fund it before launch or configure auto-reload with enough headroom. Each reload purchases more credits, so monitor actual usage separately.

What if my Gemini API requests have already stopped?

Start with the project behind the failing key. If its billing status says No credits, check that prepay setup finished and that funds reached its linked account. When a shared prepaid balance runs out, keys across its linked projects can stop together. Rotating a key on the same project leaves it on the same balance. Google’s billing status reference.

Once the balance appears, retry one small request. If it still fails, save the error message and check the project’s billing status before adding more money. If it succeeds, resume the workload and check usage after the first batch.

How big should my prepaid balance be?

Size the balance and auto-reload from your daily spend, not from the $5 minimum. Take your average daily Gemini cost over the last week, then check your busiest day. Set the auto-reload threshold above what your busiest day uses, so a reload can arrive before the balance runs out.

Plan for the price change too. Gemini 3.6 Flash and 3.8 Flash input and output rates double on January 1, 2027, and Gemini 4 Argon rates double on December 1, 2026. After the change, the same traffic uses up the same balance about twice as fast. What doubles and when.

The table below shows an illustrative workload: 4 million input tokens and 400,000 output tokens a day on Gemini 3.6 Flash, with no cached input.

At this volume, the $5 minimum lasts about a day today and about half a day after January 1. A reload threshold that covers one busy day today will cover only half of one after the increase.

Is Gemini still the cheapest model for this workload?

Switching to prepay means paying in advance, so check where the money goes before you add a larger balance. Run the same daily token counts through other models’ prices:

Token prices are only half the decision. A cheaper model may need longer prompts, produce longer answers, or fail more often on your task. Test any alternative on real requests before you move traffic or shrink your Gemini balance.

How do I check this with my own usage in Ganivra?

Ganivra’s dashboard records each Gemini call’s tokens and cost. Two views help with prepay:

• Daily spend. The Margin outlook tab shows each customer’s cost over the last seven days and their average daily cost. To check a peak, set the dashboard’s date range to a single busy day and filter by the workflow or customer that uses the affected project. Use both figures to size your balance and auto-reload threshold.
• Model comparison. This tab takes the tokens you actually used and prices them on GPT, Claude and Gemini alternatives at today’s list rates. A green verdict means your current model is cheaper; red means it costs more. You can compare per workflow, so one expensive step doesn’t hide behind the total.

Model comparison uses rates in effect on the day you open it. It won’t show the January 1 Gemini price until that date. Until then, double the Gemini 3.6 Flash or 3.8 Flash figure to estimate the new cost, and compare that with the alternatives. Tool calls, cache writes and multimodal requests are left at their recorded cost rather than repriced, and the tab shows how many calls it covered.