Standard Compute
← All providers

Gemini API Rate Limits

The Gemini API rate-limits per project across three dimensions — requests per minute (RPM), input tokens per minute (TPM), and requests per day (RPD) — plus rolling 10-minute spend caps. Exceeding any single dimension errors even if the others have headroom.

Transcribed from the official documentation · verified 2026-07-17

How Google Gemini limits work

·Tiers: Free (no billing required) → Tier 1 (billing enabled) → Tier 2 ($100 spent + 3 days) → Tier 3 ($1,000 spent + 30 days).
·Each tier carries a spend cap: Tier 1 $250, Tier 2 $2,000, Tier 3 $20,000-$100,000+.
·Separate spend-based protection: rolling 10-minute windows cap spend at $10 (Tier 1) or $200 (Tiers 2-3).
·The generous free tier is why Gemini is many agents' fallback — but daily request caps (RPD) bite long-running agents harder than per-minute limits.
·Exact per-model RPM/TPM figures are shown in the AI Studio dashboard rather than static docs.

Usage tiers

TierQualificationSpend cap
FreeActive project—
Tier 1Billing enabled$250
Tier 2$100 spent + 3 days after first payment$2,000
Tier 3$1,000 spent + 30 days after first payment$20,000-$100,000+

Plus rolling 10-minute spend windows: $10 max (Tier 1), $200 max (Tiers 2-3).

When you hit the limit

·429 RESOURCE_EXHAUSTED errors; daily (RPD) limits reset on Google's daily schedule rather than per minute.
·Hitting the rolling 10-minute spend window errors even when RPM/TPM have headroom.

Getting higher limits

·Advance tiers by enabling billing and accumulating spend + account age.
·"Request paid tier rate limit increase" via Google's form — explicitly best-effort, no guarantees.

Google Gemini rate limits — common questions

Why does Gemini rate-limit my agent even on the paid tier?

Check which dimension fired: RPD (daily request caps) catches always-on agents that per-minute limits never would, and the rolling 10-minute spend windows ($10 on Tier 1) trip on bursts of large-context calls.

Is the Gemini free tier enough to run a coding agent?

For light experimentation yes — it's the most generous free tier among the big providers. For real agent workloads the daily request caps end sessions mid-task, which is why it works better as a fallback than a primary.

Hitting these limits from an agent? The specific errors have dedicated fixes: 429 Too Many Requests — what it means & how to fix · Why your AI agent keeps getting rate limited (and how to stop it)

And if the limits themselves are the problem: this reference is maintained by Standard Compute, a flat-rate gateway with no RPM, TPM, or daily caps — heavy use is paced, never rejected. How that works →