Gemini API: Project Spend Caps and Usage Tier Revamp

Overview

Google introduced a suite of billing controls for the Gemini API on March 16, 2026, centered on two major changes: Project Spend Caps and a revamped Usage Tiers system. Both features land after sustained developer pressure following a series of billing incidents that exposed the Gemini API's lack of hard cost guardrails.

Project Spend Caps

The most immediate change is the introduction of Project Spend Caps. Developers with editor, owner, or admin roles on a project can now navigate to the Spend tab in Google AI Studio and set a monthly dollar ceiling under "Monthly spend cap > Edit spend cap." Once configured, the cap remains active until explicitly modified or disabled.

There is an important caveat: spend caps carry an approximately 10-minute enforcement delay. Google acknowledges that users are responsible for any overages accumulated during that window. For high-throughput applications this represents meaningful exposure — thousands of API calls per minute means 10 minutes of uncapped billing can be substantial — but the cap nonetheless provides a structural safeguard that previously did not exist at all.

Revamped Usage Tiers

The Usage Tiers system has been restructured to enable developers to scale capacity faster and with more transparency. Key changes include:

• Lower qualification thresholds: The spend requirements to reach higher tiers have been reduced, making it easier for developers with a solid payment history to access higher quotas.
• Automatic tier upgrades: As usage grows and payment history matures, accounts are now automatically promoted to the next tier, granting access to higher rate limits and increased monthly quota without manual intervention.
• Billing account tier caps: Each tier now carries a maximum monthly spend limit enforced at the billing account level. These caps scale upward as developers advance through tiers.

The current tier structure with spend caps is as follows:

Tier	Qualification	Spend Cap
Free	Signup	None
Tier 1	Active billing account	$250/month
Tier 2	Paid $100, 3+ days since first payment	$2,000/month
Tier 3	Paid $1,000, 30+ days since first payment	$20,000–$100,000+/month

Tier spend caps will be enforced starting April 1, 2026, with the interface already visible in AI Studio to give developers time to review and adjust their configurations.

New Billing Dashboard Tools

Alongside spend caps and tier changes, Google launched additional observability tools:

• Daily Cost Breakdown Graph in the Billing Dashboard, showing spend per project across time windows from the last 7 days to the full month, filterable by model.
• Rate Limit Dashboard providing a clear view of progress toward RPM (Requests Per Minute), TPM (Tokens Per Minute), and RPD (Requests Per Day) limits per model.

Context: A Long-Overdue Response

The absence of spending controls had become a significant pain point for the Gemini API developer community. In August 2025, a pricing configuration bug caused Gemini 2.5 Flash to incorrectly bill developers for image generation they never performed, leaving some facing charges exceeding $70,000. In February 2026, a stolen API key generated $82,314.44 in charges in under 48 hours for a developer whose regular monthly spend was around $180. Following that incident, developers broadly called for "basic guardrails against catastrophic usage anomalies."

Google's announcement makes no direct reference to these incidents, though the timing and the features introduced directly address the failure modes they exposed. All other major API providers had offered hard spending limits for some time; the Gemini API's lack of equivalent controls had been a frequently cited disadvantage for production deployments.