Google Gemini API: How to Create (and Use) an API Key

The Google Gemini API is a programming interface that allows you to use Google's AI models (e.g., Gemini 1.5 Pro or Gemini 1.5 Flash) outside of ChatGPT as well.

More and more users are working with Gemini models, and the API has become an important alternative. Current benchmarks show that Gemini 1.5 Ultra can keep up even with demanding tasks.

This is necessary, for example, if you want to create your own AI software or app, or integrate AI functions into your existing software or app.

However, having a Gemini API key is now also a must for non-developers. Because there are many apps and programs that require (optionally or mandatory) an API key. These include, for example, SheetMagic, Zapier, Make, Cursor.so or various ChatGPT alternatives.

This can save you a lot of money:

By connecting external tools to the Gemini API, you can save yourself a Gemini Advanced subscription.

In the following guide, I'll show you how to create a Google Gemini API key:

1. Create a Google account

To use the Gemini API from Google, you absolutely need a Google account (a personal account or Workspace account).

I assume you already have one. If not, you can find instructions here.

2. Sign up for Google AI Studio

AI Studio is Google's counterpart to the OpenAI Platform or Anthropic API Console.

With Google AI Studio, you can try out all Gemini models, create an API key, and even perform fine-tuning for certain models:

Open the website and sign in with your Google account.

3. Generate API key

To create a Gemini API key and use the Gemini API, click on the blue "Get API key" button at the top of the sidebar:

Now you're in the API keys overview. Click on "Create API key" here:

Next, you need to create an existing project in Google Cloud or a new project (every API key from Google must be linked to a project in Google Cloud).

Click on "Create API key in new project" here:

After a bit of waiting, your API key will be displayed. Click the "Copy" button:

4. What does the Google Gemini API cost?

The Google Gemini API can be used either in the free tier or with "Pay-as-you-go".

The free tier is more than enough for occasional or private use. The limits for Gemini 1.5 Flash, for example, are very generous:

Important: The free tier is not available in Germany. If you try to activate it, an error message will be displayed (see below). However, it works if you're logged in via a VPN in the USA (e.g., via NordVPN).

With Gemini 1.5 Pro, however, you only have two requests and 32,000 tokens per minute and 50 requests per day. This means it's not possible to make requests with more extensive documents of over 16,000 words:

With Pay-as-you-go, you pay according to the amount of tokens you send to the Gemini models (input) and receive as a response (output). One token corresponds to approximately 4 characters, or about half a word. One million tokens therefore corresponds to 500,000 words.

For Gemini 1.5 Flash, for example, one million tokens costs $0.35 for input and $1.05 for output.

5. Set up Pay-as-you-go

If you want to use the Google Gemini API with higher limits in the "Pay-as-you-go" plan, you need to add your payment details in Google Cloud.

To do this, click on the "Set up billing" link in the API overview in Google AI Studio:

You will then be redirected to the Google Cloud Console, where you need to link a billing account with your project:

6. Test API key

If you want to test whether your API key works, you can do this very easily with the following cURL command (e.g., in Mac Terminal):

curl \
  -H 'Content-Type: application/json' \
  -d '{"contents":[{"parts":[{"text":"Explain how AI works"}]}]}' \
  -X POST 'https://generativelanguage.googleapis.com/v1beta/models/gemini-1.5-flash-latest:generateContent?key=YOUR_API_KEY'