Google AI Studio rate limits and quotas: RPM, TPM, daily limits, spend caps, tiers, and 429 errors
Updated: Sep 17
Google AI Studio does not have one universal daily limit. Gemini API capacity is enforced through a combination of requests per minute, tokens per minute, requests or tokens per day, and spend-based controls, and the exact values depend on the model and usage tier.
Google explicitly says active quotas should be checked in AI Studio because published limits can vary over time and preview or experimental models often have tighter ceilings.
··········
[[ADPLUS_300x250_1]]
RPM, TPM, AND DAILY QUOTAS MEASURE DIFFERENT BOTTLENECKS
........
Limit | What it measures | Typical failure mode |
|---|---|---|
RPM | Requests per minute | Too many calls in a short interval |
TPM | Input/output token throughput per minute | Requests are too large or too frequent |
RPD / TPD | Requests or tokens per day where a model uses daily caps | Daily allocation exhausted |
Spend rate | Dollar-value usage over a rolling 10-minute window | Paid workload is generating cost too quickly |
........
··········
THE FREE TIER IS MODEL-SPECIFIC
Free-tier rate limits are not a single account-wide number. Each eligible model can have its own RPM, TPM, daily, or other quota. Preview and experimental models may have significantly lower limits than stable models.
··········
PAID TIERS INCREASE CAPACITY AS BILLING HISTORY GROWS
Tier 1 starts after linking an active billing account. Tier 2 requires at least $100 paid plus three days from the first successful payment, while Tier 3 requires at least $1,000 paid plus 30 days. The project receives higher quota as the linked billing account moves through these levels.
··········
SPEND-BASED LIMITS ADD ANOTHER CEILING
Google currently documents rolling 10-minute spend-rate limits of $10 for Tier 1, $50 for Tier 2, and $200 for Tier 3 where those controls apply. These limits are separate from RPM and TPM.
··········
429 RESOURCE_EXHAUSTED DOES NOT ALWAYS MEAN THE SAME THING
A 429 response can indicate that RPM, TPM, a daily quota, or a spend-rate control has been reached. The correct response depends on the cause: wait and retry, reduce request size, slow concurrency, move to a different model, or qualify for a higher usage tier.
··········
WHY COPYING SOMEONE ELSE'S QUOTA TABLE CAN BE MISLEADING
Google notes that listed limits are not guaranteed and actual capacity can vary with account status, tier, and model lifecycle. A quota screenshot from another account therefore does not prove the same allowance applies to your project.
··········
HOW TO CHECK THE LIMIT THAT ACTUALLY APPLIES
Use Google AI Studio's project and usage views to inspect the project's active quotas and system limits. For production planning, record the model ID, billing tier, active RPM/TPM ceilings, average tokens per request, and peak concurrency rather than relying on a generic 'daily limit' number.
··········
FOLLOW US FOR MORE
··········
DATA STUDIOS
··········
