Gemini API free-tier limits: the short answer
There is no single request allowance that applies to every Gemini API model and project. Google says API limits vary by model, are measured across limits such as requests per minute (RPM), input tokens per minute (TPM), and requests per day (RPD), and are applied per project—not per API key. To see the limits currently assigned to your project, use the “View your active rate limits” link in Google AI Studio. Google’s public rate-limits page describes how quotas work, but it does not provide one fixed, universal free-tier number for every model. (ai.google.dev)
The free tier provides free input and output tokens for certain models, subject to their free-tier limits. It is not the same as a Gemini consumer-app plan, a Google AI subscription, or a Vertex AI/Gemini Enterprise Agent Platform quota. For production budgeting, confirm both your project’s active model limits and the current paid price for the model you plan to use; free-tier limits are not a capacity guarantee. (aistudio.google.com)
What the limits measure
| Limit | What it measures | What to keep in mind |
|---|---|---|
| RPM | Requests per minute | Exceeding this limit can cause a rate-limit error even if other limits are not reached. |
| TPM | Input tokens per minute | A request may fit the RPM limit but exceed the token limit. |
| RPD | Requests per day | Google says the daily quota resets at midnight Pacific time. |
| Other model or tool limits | For example, image-generation or built-in-tool usage | These apply only where relevant; check the model and tool limits that appear for your setup. |
Google says the applicable limits depend on the model, and experimental or preview models may have more restrictive limits. That is why a quota figure copied from another model, project, or older guide can be misleading. (ai.google.dev)
Which Gemini product are you using?
The phrase “Gemini free tier” can refer to different products, but their allowances are not interchangeable:
- Gemini Developer API / Google AI Studio API: The free API tier offers free input and output tokens for selected models, within model-specific limits. Check the active rate limits for your project in AI Studio. (
aistudio.google.com) - Gemini consumer app or Google AI subscription: These have their own product features and usage allowances. Google says subscription benefits for developer use apply within the AI Studio web interface; direct API-key use is billed and managed separately. (
ai.google.dev) - Vertex AI or Gemini Enterprise Agent Platform: These use Google Cloud platform quotas and billing arrangements. Do not substitute their quota tables for an AI Studio free-tier limit. Google’s Cloud quotas documentation is explicitly about the Agent Platform, not the Gemini Developer API free tier. (
ai.google.dev)
In particular, consumer-app prompt or image allowances do not tell you how many API requests your application can make.
What does Gemini API access cost after the free tier?
Google separates free-tier access from paid API usage. Its pricing page describes the free tier as limited access to selected models with free input and output tokens; paid access is positioned for applications needing higher rate limits and more features. The price you pay depends on the model and the amount and type of tokens processed, so check Google’s live pricing table before setting a budget. (aistudio.google.com)
As one dated pricing example, Google announced Gemini 3.1 Flash-Lite at $0.25 per 1 million input tokens and $1.50 per 1 million output tokens on March 3, 2026. At those published rates, a hypothetical workload of 10 million input tokens and 2 million output tokens would cost $5.50 before any separately priced tools or other charges: (10 × $0.25) + (2 × $1.50). Treat this as an illustration, not a guaranteed current quote; confirm the model’s current price on Google’s pricing page before relying on it. (blog.google)
Google’s billing documentation says moving a project from the Free Tier to a Paid Tier requires linking billing and prepaying at least $5, or the equivalent in another currency. That is a minimum prepaid balance—not a monthly subscription price. Paid API usage is billed separately from Google AI subscription plans. Do not assume that hitting a free quota automatically switches a request to paid usage; check the project’s billing configuration and usage before deployment. (ai.google.dev)
There is also a data-use trade-off: Google’s API pricing page says content from Free Tier usage may be used to improve Google products, while Paid Tier content is not used for that purpose under the stated terms. Review Google’s current terms and your organization’s data requirements before sending sensitive production data. (aistudio.google.com)
How to decide whether the free tier fits
Use this quick check before building around free access:
- Identify the exact channel and model. Confirm that you mean Gemini Developer API/AI Studio, not the consumer app or another Google Cloud service.
- Check the active quota for your project. Open the active-rate-limits view in Google AI Studio. Limits are per project, not per API key, and vary by model. (
ai.google.dev) - Estimate both request and token demand. Compare expected requests per minute, input tokens per minute, and daily requests with the limits shown for that model. Leave room for traffic spikes and retries rather than planning to run continuously at the ceiling.
- Price the workload that may need paid usage. Estimate input and output tokens separately using the current model pricing. Include any tools or features that may have separate charges. (
ai.google.dev) - Decide whether the free tier is an acceptable production dependency. Google describes free access as limited to selected models and presents paid access as the option for production applications needing higher volumes. The documentation does not make free quotas a production-capacity guarantee. (
aistudio.google.com)
If a request returns 429 RESOURCE_EXHAUSTED, first check which limit or billing condition applies rather than assuming the daily request allowance is the cause. Google recommends exponential backoff for errors that are appropriate to retry; its troubleshooting guidance also cautions against retrying client errors such as 400, 402, or 403. (ai.google.dev)
Bottom line
Gemini API free access can be useful for trying selected models, but there is no dependable one-number answer for every developer: the applicable limits vary by model and project. Check your project’s active limits in AI Studio, keep API quotas separate from consumer Gemini allowances, and use Google’s current API pricing—not subscription prices or third-party quota tables—for production budgeting.
Sources
Rate limits | Gemini API | Google AI for Developers
Gemini Developer API pricing | Google AI for Developers
Billing | Gemini API | Google AI for Developers
Google AI plans | Gemini API | Google AI for Developers
Gemini 3.1 Flash-Lite: Built for intelligence at scale
Troubleshooting guide | Gemini API | Google AI for Developers