Gemini API Free Tier (Google AI Studio): Free Models & Rate Limits (2026)

Short answer: yes, the Gemini API is free to start. Google's official pricing page lists a Free tier for most Flash, Flash-Lite, Live, TTS and embedding models, plus Gemini 2.5 Pro and the open Gemma 4 models. You need a Google account and an API key from Google AI Studio, no card. The trade-offs are per-project rate limits that Google does not publish as fixed numbers, and the fact that free-tier prompts and responses can be used to improve Google's products. This page covers what is free, what is not, how the limits work, and when it makes sense to pay.

What the Gemini API free tier includes

On the pricing page every model has a 'Free tier' column that reads either 'Free of charge' or 'Not available'. As of 17 October 2026 the free models fall into a few groups:

Gemini 2.5 Pro prices quoted above are for prompts up to 200k tokens; longer prompts cost $2.50 input and $15.00 output per 1M tokens on the paid tier. On the free tier you pay nothing for either, but you are capped by rate limits instead of a bill.

What is not free

• Gemini 4 Argon — the pricing page states it 'is not available on the Free tier'.
• Gemini 3.1 Pro Preview and Gemini Omni Flash (Omni Flash is described as generally available 'on the paid tier').
• Image generation: Nano Banana 2.1, Gemini 3 Pro Image (Nano Banana Pro), Gemini 3.1 Flash Image and 3.1 Flash Lite Image. Gemini 2.5 Flash Image is scheduled to shut down on 2 October 2026.
• Video and music: Veo 3.1 (Standard, Fast and Lite) and Lyria.
• Gemini 2.5 Pro Preview TTS.
• Batch API, Flex and Priority processing are paid-tier features; the Batch API runs at roughly 50% of Standard prices.

Grounding with Google Search has its own allowance. The pricing page lists 5,000 free search requests per month shared across Gemini 3 and newer models on the paid tier (then $14 per 1,000), and for Gemini 2.5 Flash and Flash-Lite the free tier shows grounding as 'Free of charge, up to 500 RPD'.

Who can use it

• Age: the Gemini API terms require you to be 18 or older. AI Studio shows an error if you do not meet the minimum age.
• Location: the API and AI Studio work only in the countries and territories on Google's available regions list. The list includes the US, UK, Switzerland and EU countries such as Germany, France, Italy and Spain; mainland China and Russia are not on it.
• EEA, Switzerland and UK apps: the terms say you may use only Paid Services when making an API client available to users in the European Economic Area, Switzerland or the UK. Testing on your own is fine; shipping to users there means enabling billing.
• Business use: the terms describe the APIs as being for professional or business purposes, not consumer use.
• Students: there is no separate student API tier. Students who want the consumer Gemini app with more features should look at Google AI Pro free for students, which is a different product from the API.

How to get a free Gemini API key (step by step)

1. Go to Google AI Studio and sign in with your Google account. Accept the Terms of Service; Google creates a default Cloud project and key for new users.
2. Open the API Keys page. Keys marked 'Standard' in the Key Type column can be used; click Create API key to make a new one.
3. If you already use Google Cloud, open Dashboard → Projects → Import projects to bring an existing project into AI Studio first. AI Studio shows up to 10 projects at a time.
4. Optionally tick Restrict to Gemini API only for the key. Google has rejected unrestricted standard keys and blocked unrestricted dormant keys since 7 May 2026.
5. Set the key as `GEMINI_API_KEY` (or `GOOGLE_API_KEY`, which takes precedence if both exist). The official SDKs pick it up automatically.
6. Open the rate-limit page to see your per-model limits before you build anything that depends on throughput.

Our Gemini API key guide walks through the same flow in more detail, including how to test the key from the command line.

export GEMINI_API_KEY="your-key"
curl "https://generativelanguage.googleapis.com/v1beta/models/gemini-2.5-flash:generateContent" \
  -H "x-goog-api-key: $GEMINI_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"contents":[{"parts":[{"text":"Explain rate limits in one sentence"}]}]}'

How the free-tier rate limits work

Google's rate-limits page describes the mechanics but not fixed free-tier numbers. Here is what is officially documented:

• Limits are measured in requests per minute (RPM), input tokens per minute (TPM) and requests per day (RPD). Some models also have a tokens-per-day limit, and image models have images per minute.
• Hitting any single dimension triggers an error, even if the others have room. Google's example: with a 20 RPM limit, the 21st request in a minute fails.
• Limits are per project, not per API key. Ten keys in one project share one quota.
• Daily quotas reset at midnight Pacific time.
• Experimental and preview models have stricter limits than stable ones.
• Google says specified limits 'are not guaranteed and actual capacity may vary'.

Exact free-tier numbers for your project are on the AI Studio rate-limit page. Figures you may see on blogs (such as 10 RPM or 250 requests per day for a given model) are not printed in Google's docs and may not match your account.

Usage tiers above Free

Moving from Free to Tier 1 usually takes effect instantly; later upgrades take up to about 10 minutes and are automatic once you qualify. Paid tiers also have a rolling 10-minute spend limit ($10 on Tier 1, $50 on Tier 2, $200 on Tier 3) that returns `429 RESOURCE_EXHAUSTED` when hit.

Privacy: the main cost of 'free'

The pricing page has a row called 'Used to improve our products'. For the free tier it says Yes; for the paid tier it says No. The Gemini API terms add that, for Unpaid Services, human reviewers may read, annotate and process API input and output, after Google disconnects it from your account, key and project. Google's own advice is not to submit sensitive, confidential or personal information to the Unpaid Services.

Two practical consequences follow. First, do not send customer data, private code or anything under an NDA through a free-tier project. Second, once at least one project on your account has billing enabled, Google says your AI Studio prompts are treated under the Paid Services terms, so a small prepaid balance can change how your data is handled.

Moving to the paid tier and what it costs

Per the billing docs, upgrading means linking a billing account and prepaying at least $5. Click Set up billing on the API Keys or Projects page in AI Studio. New users default to the Prepay plan: usage is deducted from a balance, and when it reaches $0 every key in every project on that billing account stops working at once. Eligible existing accounts can stay on Postpay, but switching from Prepay to Postpay is not supported. You can also set experimental project-level spend caps in AI Studio, with up to about 10 minutes of billing latency.

Worked cost examples (Standard paid tier)

Running the same jobs through the Batch API roughly halves these figures, at the cost of asynchronous results. For the full plan-by-plan breakdown see our Gemini API pricing page.

Google Cloud credits do not cover the Gemini API

A common assumption is that the $300 Google Cloud Free Trial can pay for Gemini API usage. It cannot. Google's Gemini billing docs state that free trial credit can't be used toward the Gemini API or AI Studio, and that starting March 2026 Gemini API costs are specifically excluded from the $300 Free Trial. The Google Cloud Free Program docs repeat this and point to the Gemini free tier or a Prepay plan instead. Credits granted before the change can still be spent until they expire. Details of what the trial does cover are on our Google Cloud $300 free trial page. Startups with larger needs should look at the Google for Startups Cloud Program.

Getting the most out of the free tier

• Pick the lightest model that works. Flash-Lite models are built for high-volume, low-latency jobs such as classification, extraction and routing. Reserve 2.5 Pro for tasks that need it, since preview and larger models tend to have tighter limits.
• Use one project per app. Because limits are per project, separating a hobby bot from a prototype keeps one from starving the other.
• Handle 429s gracefully. Add exponential backoff and queue work so a burst does not fail outright.
• Keep prompts short. TPM counts input tokens, so trimming long system prompts or chat history directly raises how many requests fit in a minute.
• Protect the key. Google says to treat it like a password, never commit it to source control, and never ship it in a web or mobile client. Put a small backend proxy in front.
• Embed with Gemini Embedding 2 for small retrieval projects; it is free on the free tier and $0.20 per 1M text tokens when paid.

How it compares with other free AI APIs

Gemini's free tier is unusual because it includes a Pro-class model (2.5 Pro) and multimodal input at no cost. The catch is data use. Other options trade off differently:

• Groq API free tier publishes exact per-model limits and is very fast, but only for open-weight models such as gpt-oss.
• Mistral API free tier is good for experimenting with several model families from one key.
• OpenRouter free models give one API across many providers' free endpoints.
• DeepSeek has no confirmed free credit but very low per-token prices.

Our roundup of free AI API keys and the broader guide on how to get free AI credits compare these side by side.

Recent changes worth knowing

• March 2026: Gemini API costs excluded from the Google Cloud $300 Free Trial (Gemini billing docs).
• 7 May 2026: unrestricted dormant API keys blocked; unrestricted standard keys are rejected (API key docs).
• 2 October 2026: Gemini 2.5 Flash Image scheduled shutdown (pricing page).
• 1 December 2026: Gemini 4 Argon paid prices scheduled to rise from $2.00 to $4.00 input and $10.00 to $20.00 output per 1M tokens.
• 1 January 2027: Gemini 3.8 Flash, 3.6 Flash, 3.8 TTS models and Robotics ER 2 Preview paid prices scheduled to double. Free-tier access for these models is not affected by the listed change.

Some third-party sites reported that Gemini 2.5 Pro was removed from the free tier in 2026. Google's official pricing page still lists Gemini 2.5 Pro as 'Free of charge' on 17 October 2026. If your rate-limit page shows no quota for a model, that page is the authority for your project.