Google Slashed Gemini API Quotas 92% and Nobody Told You — The Full Antigravity Timeline
If you've been building on the Gemini API and noticed things getting tighter over the past few months, you weren't imagining it. A detailed breakdown published by Apiyi.com has catalogued the full timeline of Google AI Studio's rolling quota cuts from December 2025 through March 2026 — and the numbers are striking. Free-tier requests per day (RPD) dropped 92%, falling from 250 to just 20. Requests per minute (RPM) were cut by 50%. And in March 2026, Google quietly introduced an AI Credit system, where even $25 only buys 2,500 credits. Perhaps most alarming: paid Ultra subscribers reported unannounced throttling, with no official communication explaining the change.
The shift reflects a broader strategic pivot at Google AI Studio, where visibility into quota limits has moved away from a public documentation table and into per-project settings inside the AI Studio dashboard itself — meaning developers who aren't actively monitoring their usage dashboards can miss critical changes entirely. For teams in the EU, EEA, Switzerland, and the UK, there's an additional layer of complexity: Google's terms prohibit use of the free tier for user-facing production applications in those regions. The article includes actionable workarounds and alternative routing strategies for developers who need to stay on Gemini without budget surprises.
The practical upshot for anyone building production applications on Google's AI infrastructure: treat free-tier access as evaluation-only, budget for paid tiers with a buffer, and monitor your AI Studio dashboard proactively. The hidden constraint landscape for Gemini API has fundamentally changed since 2025, and the documentation hasn't always kept pace.