How to Get a Free Gemini API Key for Cereal's AI Browser Agent

TL;DR: Sign in at Google AI Studio, click Create API key and copy it. In Cereal, open Settings → AI, pick OpenAI-compatible, set the base URL to https://generativelanguage.googleapis.com/v1beta/openai, paste the key, choose a Gemini Flash model and click Test connection. The AI Browser Agent now runs on Gemini’s free tier.

Cereal’s AI Browser Agent drives your own Chrome to reach a goal you describe in plain language: check whether something is in stock, read a price, look up an order. It needs an AI model to think with, and you bring that model yourself. If you don’t want to pay for one, a free Gemini API key from Google is the quickest way to start.

This guide takes about five minutes. You need a Google account and the Cereal desktop app.

Why Gemini’s free tier

Cereal is free, and you pay your AI provider directly. Google’s Gemini API has a free tier that covers its Flash and Flash-Lite models, which are fast, cheap and good at the step-by-step work a browser agent does. On the free tier you pay nothing at all; Google limits how many requests you can make per minute and per day instead.

One thing to know before you start: on the free tier, Google may use what you send to improve its products. That includes the pages the agent reads. For tasks on sites where you’re signed in, or anything private, use a paid key or a local model instead. Nothing passes through Cereal either way: your key and pages go only to the provider you pick.

Step 1: Create the key in Google AI Studio

Go to aistudio.google.com/apikey and sign in with your Google account.
If this is your first visit, accept the terms. Google creates a default project for you; every key belongs to a project.
On the API Keys page, click Create API key.
Copy the key and keep it somewhere safe, such as your password manager.

Treat the key like a password. Don’t paste it into chats, screenshots or shared files. If it ever leaks, create a new one in AI Studio and delete the old one.

If the Create API key button is greyed out, your Google account probably belongs to a work organization that limits it. Use a personal Google account, or create a new project that isn’t tied to an organization.

Step 2: Add the key to Cereal

The key goes into the Cereal app, not into the script. Scripts never see it.

Open Cereal and go to Settings → AI.
Set Provider to OpenAI-compatible. Gemini offers an OpenAI-compatible endpoint, so this is the one to pick.

Set Base URL to:

https://generativelanguage.googleapis.com/v1beta/openai

Paste your key into API key.
Click Show models and pick a Gemini Flash model, such as Gemini 3.8 Flash. Flash-Lite models are cheaper and usually get higher free-tier limits, but make more mistakes on tricky pages.
Click Test connection, then Save.

If the test fails with an authentication error, the key was copied with a missing character or an extra space. Copy it again from AI Studio.

Step 3: Run your first task

The AI Browser Agent is already in Cereal: it installs automatically when you sign in, so there’s nothing to add. Open it, then:

Write a goal and a start URL, for example: tell me the price of the cheapest hammer on a hardware store’s site.
Pick Once to get an answer, or Repeat to check on an interval until a condition holds, like an item coming back in stock.
Start the task. A Chrome window opens and you can watch the agent work.

The agent asks for your OK before it buys, pays, deletes or sends anything. Repeat tasks learn the steps on the first run and replay them later without calling the model, so routine checks use almost none of your free requests.

When you hit the free tier’s limits

The free tier caps requests per minute and per day. When you hit a limit, the run stops with a rate-limit error. You can wait and try again, switch to a Flash-Lite model, which usually has higher limits, or turn on billing for your Google project to move to the paid tier.

Once billing is on, you pay per token. Cereal’s own limits stop every run at $2 and every task at $5 by default, and you can change them in the task. Set a budget alert in your Google Cloud billing settings as well.

Next steps

You now have a free model connected and the agent ready to go. Some ideas to try:

A Repeat task that tells you when a sold-out product is back.
A Once task that reads a value you check every morning.
A signed-in lookup, with your login saved as a sign-in secret so the AI never sees the password. Use a paid key or local model for this one.

Questions or stuck on a step? Ask in our Discord.