October 21, 2026. A Gemini API project on the free tier can send Gemini 3.8 Flash 5 requests a minute and 20 a day, according to the rate limit page Google AI Studio showed for our own free project today. The newer Flash-Lite models get 15 a minute and 500 a day, while Gemini 4 Argon, Gemini 3.1 Pro and Gemini 2.5 Pro get none. Google no longer prints these figures: its rate limits page, updated October 15, points to AI Studio and says specified limits are not guaranteed.
Key numbers
| Item | Number |
|---|---|
| Gemini 3.8 Flash, free tier (250,000 input tokens a minute) | 5 RPM, 20 a day |
| Gemini 3.5 and 3.1 Flash-Lite, free tier | 15 RPM, 500 a day |
| Gemini 4 Argon, 3.1 Pro and 2.5 Pro, free tier | 0 requests |
| Gemma 4, free tier (30 RPM) | 14,400 a day |
| Daily limit reset (per project, not per key) | midnight Pacific |
| Paid Tier 1 entry (linked billing account) | $5 minimum prepay |
| Gemini 3.8 Flash, Tier 1 (2 million TPM) | 1,000 RPM, 10,000 a day |
| Gemini 3.8 Flash, paid price per 1M tokens (doubles January 1, 2027) | $0.75 in, $3.75 out |
| Grounding with Google Search, paid (then $14 per 1,000; not on free) | 5,000 free a month |
| Prepaid credits (non-refundable) | expire after 12 months |
Free tier and Tier 1 limits read on 21 October 2026 in Google AI Studio for one free project; prices and billing rules from Google's Gemini API pricing, rate limits and billing pages, read the same day.
Free tier limits by model, read in AI Studio
Requests per minute (RPM), input tokens per minute (TPM) and requests per day (RPD), read October 21:
- Gemini 3.8 Flash, 3.6 Flash, 3 Flash and 2.5 Flash: 5 RPM, 250,000 TPM, 20 RPD.
- Gemini 3.5 Flash-Lite and 3.1 Flash-Lite: 15 RPM, 250,000 TPM, 500 RPD. Gemini 2.5 Flash-Lite gets 10 RPM and 20 RPD.
- Gemini 4 Argon, Gemini 3.1 Pro and Gemini 2.5 Pro: 0 on every counter. Google's pricing page, updated October 19, labels Argon not available on the free tier but still marks Gemini 2.5 Pro free of charge, so check your own project.
- Gemma 4 and embeddings: Gemma 4 gets 14,400 RPD at 30 RPM; Gemini Embedding 2 gets 1,000 RPD. Image, video and music models get 0.
Limits apply per project, not per API key, so a second key shares the same 20 requests, and daily counters reset at midnight Pacific time. Some third-party guides still quote about 10 requests a minute and 1,500 a day for free Flash; our project shows 5 and 20.
What free costs besides requests
The pricing page marks free tier content as used to improve Google's products; paid tier content is not. Grounding with Google Search is not available on the free tier for Gemini 3 models; paid projects get 5,000 grounded requests a month, shared across Gemini 3 and newer models, then $14 per 1,000. An invite-only Contributor tier sells the same data trade at half the Standard price.
Leaving the free tier: $5 and a billing account
Tier 1 needs a linked billing account and a prepaid top-up of at least $5, per Google's billing page. AI Studio's tier comparison puts Gemini 3.8 Flash on Tier 1 at 1,000 RPM, 2 million TPM and 10,000 RPD, and the newer Flash-Lite models at 4,000 RPM and 150,000 RPD. Tier 1 carries a $250 billing cap; Tier 2 follows after $100 paid and 3 days. Prepaid credits expire after 12 months and are non-refundable, every key on the billing account stops with an HTTP 402 error when the balance reaches $0, and Google Cloud's $300 welcome credit cannot be spent on the Gemini API.
Gemini 3.8 Flash is $0.75 per million input tokens and $3.75 per million output through December 31, rising to $1.50 and $7.50 on January 1, 2027, the change in our Gemini Flash price increase brief. Gemini 3.1 Flash-Lite is $0.25 and $1.50.
What it means for operators
Take a Shopify store rewriting 2,000 product descriptions. On the free tier, Gemini 3.8 Flash at 20 requests a day needs 100 days; Flash-Lite at 500 a day needs four. On the paid tier, at about 800 input and 400 output tokens a product, Gemini 3.8 Flash does the job for about $4.20, $1.20 of input and $3.00 of output, which is less than the $5 minimum top-up, or $8.40 after January 1.
- Use the free tier for tests and public text such as product copy, on Flash-Lite, with 25 times the daily allowance.
- Keep customer data off free keys. Emails, call transcripts and CRM records belong in a paid project.
- Read your own limits on AI Studio's rate limit page before scheduling around them.
- Watch the prepaid balance. At $0, every key on the account stops.
Tracking how Gemini describes your brand takes more than 20 requests a day and, on Gemini 3 models, grounded answers, which means a paid project. That measurement is part of our AI search optimization work, and catalog jobs sit in our e-commerce services. For other platforms, see our Claude API rate limits and OpenAI API usage tiers briefs.
Frequently Asked Questions
On a free project in Google AI Studio on October 21, 2026: Gemini 3.8 Flash, 3.6 Flash, 3 Flash and 2.5 Flash at 5 requests a minute, 250,000 input tokens a minute and 20 requests a day; Gemini 3.5 and 3.1 Flash-Lite at 15 a minute and 500 a day; Gemini 4 Argon, 3.1 Pro and 2.5 Pro at 0. Google says limits vary by project and are not guaranteed.
Requests per day reset at midnight Pacific time. Limits are counted per project, not per API key, so creating a second key on the same project does not add requests.
Yes. Google's Gemini API pricing page marks free tier content as used to improve Google's products, and paid tier content as not used. Keep customer data such as emails, call transcripts and CRM records on a paid project.
Link a billing account in Google AI Studio and prepay at least $5 to reach Tier 1, where AI Studio lists Gemini 3.8 Flash at 1,000 requests a minute and 10,000 a day. Tier 2 follows after $100 of payments and 3 days.