Source: Gemini API — Rate limits
Gemini Developer API limits: RPM/TPM/RPD per project (not per key), midnight-Pacific RPD reset, spend-based usage tiers, a rolling 10-minute spend limit, Priority inference limits, and Batch enqueued-token caps.
Contents
- What: official API documentation
- Publisher: Google
- Date: living docs, read 2026-11-14
- Original: aistudio.google.com/docs/rate-limits (formerly ai.google.dev/gemini-api/docs/rate-limits)
Summary
How the Gemini Developer API (AI Studio) meters usage. Vertex AI uses a different model; see Vertex AI Standard PayGo.
Key claims
- Main dimensions are RPM, TPM (input) and RPD; exceeding any one returns an error. IPM applies to image models; some models have TPD. Built-in tools have their own limits (Search grounding on Gemini 3+: 1,500 RPD per project by default).
- Limits apply per project, not per API key; RPD resets at midnight Pacific.
- Tiers: Free; Tier 1 (linked billing); Tier 2 ($100 cumulative spend and 3 days since first payment); Tier 3 ($1,000 and 30 days). Upgrades take effect within about 10 minutes.
- A spend-based limit over a rolling 10-minute window (Tier 1 $10, Tiers 2–3 $200) returns
429 RESOURCE_EXHAUSTED. - Priority inference has its own limits, by default 0.3x the standard limit, and also counts toward interactive limits.
- Batch API: 100 concurrent batch requests, 2 GB input file, 20 GB storage, and per-model enqueued-token caps by tier.
- Published limits are not guaranteed; increases go through a request form.