Google AI Studio (2026): What You Can Do There and Who Pays

Google's own quickstart sums it up in two lines. AI Studio lets you quickly try out models and experiment with different prompts, and when you're ready to build, you select Get code, pick your programming language and carry on with the Gemini API. It's the developer side of Gemini. If you just want to chat with Gemini on your phone, that's the consumer app, and I cover it in my guide to using the Gemini app.

AI Studio, Build mode and the Gemini app side by side

People land on AI Studio expecting a chatbot and find a control panel. Here's how the pieces line up, as of Nov 2026:

	Playground	Build mode	Gemini app
What you do there	Write a prompt, test it, tweak settings	Describe an app and get a working one with a live preview	Chat with a general-purpose assistant
Main controls	Run settings: model parameters, safety settings, tools	A chat panel, a Code tab and annotation mode	The conversation itself
What you take away	Code from the Get code button, or a saved prompt	A web or Android app you can share or deploy	Answers and drafts
Who it suits	Anyone testing a prompt before building	People prototyping a small tool	Everyday use

AI Studio is also not the same thing as a Gemini subscription. Google AI Pro is the consumer plan for the Gemini app; AI Studio billing runs through a Google billing account and prepaid credits, which I explain below.

Your first chatbot in the Playground, step by step

Google's quickstart walks through a fun example: a friendly chatbot that talks as if it's an alien living on Europa, one of Jupiter's moons. Swap the alien for your own customer service bot and the steps are identical.

1. Open Google AI Studio. The Playground opens by default with a new chat prompt.
2. Click Run settings in the top-right corner to expand the panel, and find the System Instructions field.
3. Paste one line that tells the model who it's meant to be. Google's version reads: You are an alien that lives on Europa, one of Jupiter's moons.
4. In the box labelled Type something, ask what a user might ask, such as What's the weather like?, and click Run.
5. Read the reply, then tighten the instructions. Google's second version names the bot Tim, tells it to keep answers under 3 paragraphs and asks for an upbeat, chipper tone.
6. When it behaves, click Get code to start coding, or save the prompt to work on later and share with others.

System instructions: the one box that changes the bot

That example teaches the most useful thing in AI Studio. One line in System Instructions turned a general model into a character. Google admits a single instruction isn't always enough: without more detail, the weather answer tends to be very long and can take on a mind of its own. Adding a length limit and a tone fixed it. You can edit system instructions after a chat has started, so you can keep adjusting and re-running the same question until the answers sound right.

Google's own business example is a customer service chatbot that only talks about a company's product. That's the test I'd run first: write the rules you'd give a new member of staff, paste them in, then throw awkward customer questions at it.

Two more things sit in the Run settings panel. You can switch on tools such as structured output, function calling, code execution and grounding, and you can change safety settings. And Chat prompts aren't the only kind: AI Studio also offers Realtime streaming, Video gen and more.

One catch from Google's notes: every message between you and the model is included in the prompt, so a long conversation keeps growing. Eventually you may hit the model's token limit, which is the maximum length of text it can accept. Start a fresh chat prompt when a test session gets long.

Build mode: an app from a single description

Build mode is what Google calls vibe coding. You describe what you want and AI Studio generates the code and files, with a live preview of your app on the right. There are four ways in:

• A prompt: type a description, add AI Chips for features such as image generation or Google Maps data, or use the speech-to-text button.
• I'm Feeling Lucky: Gemini writes a project idea for you.
• The App Gallery: open a project and select Copy App to remix it.
• GitHub: import existing code from the Add files menu.

Web apps get a React front end by default and a Node.js server behind it. You can pick a native Android app instead, built in Kotlin and Jetpack Compose, which you preview in a browser-based emulator and can publish to the Play Store for internal testing. Google says the Antigravity Agent manages multiple files and verifies code updates to reduce hallucinations. On top of that you get npm packages, a Secrets area for keys, Multiplayer, Firebase Firestore and Sign in with Google, and connections to Gmail, Sheets, Docs, Drive and Calendar, with AI Studio handling the OAuth setup. Annotation mode lets you highlight part of your app's screen and describe the change you want.

Whose quota a shared app spends

This is the warning I'd want every business owner to read. When you build a web app that uses the Gemini API, AI Studio sets up your GEMINI_API_KEY as a server-side secret, so it never appears in the browser and people using your app can't see it. Good. But when you share the app, their API calls count toward your usage limits, and if you use paid models, costs may apply. AI Studio gives you a heads-up before you share if your app could run up costs.

Apps are private by default. Anyone you share with can see the code and fork it, and if you give edit permission, they can change it. Deploy to Cloud Run and the public app uses your API key for every user's Gemini calls, and Cloud Run pricing may apply on top. Download it as a ZIP instead and you'll set the GEMINI_API_KEY environment variable yourself wherever you host it. Google also reminds app owners that they're responsible for the app's behaviour and all the data it handles.

Free Tier, paid tiers and the prepay rule

New accounts begin on the Free Tier, which gives access to certain models up to their free tier rate limits. I keep the detail on those limits in my breakdown of what the Gemini API gives you free. To move up, you link a billing account and prepay a minimum of $5 in credits. As of Nov 2026 Google's billing page lists these tiers:

Tier	How you qualify	Billing tier cap
Free	Active project or free trial	None listed
Tier 1	Set up and link an active billing account	$250
Tier 2	Paid $100 plus 3 days from first successful payment	$2,000
Tier 3	Paid $1,000 plus 30 days from first successful payment	$20,000 to $100,000 and up

Here's the honest catch. Google lists keeping your prompts and responses out of Google product improvement as one of the reasons to move to a paid tier. So on the Free Tier, assume your prompts may be used to improve Google products, and keep customer data out of it.

The prepay rules matter too. The Gemini API only serves requests while you have a positive Prepay credit balance. When it hits zero, every API key in every project linked to that billing account stops at once, with an HTTP 402 Payment Required error. Unused credits expire after 12 months and are non-refundable, except after switching to a Postpay account. Billing runs about 10 minutes behind, so long jobs can overshoot your balance. You can set auto-reload with a monthly auto-charge limit, and track spend under Dashboard, then Usage.

Who I'd send to AI Studio, and who should skip it

I'd use AI Studio to test a prompt before paying anyone to build it. A short session in the Playground with your own system instructions tells you whether an idea works, and it costs nothing on the Free Tier. Skip Build mode for anything that touches customer data until you've read the sharing and key notes above, and you've decided who pays when other people use it. If you'd like to compare free keys from other vendors first, I list them in my roundup of free AI API keys.

If you'd like to know whether building a small app with AI is worth it for your business, that's the kind of AI enablement I help with. I run AI training and workshops for teams, and help businesses with AI enablement and workflow automation, so the busywork runs itself and people get time back. Most of that work is with SaaS marketing teams and small businesses. Here's how we can work together.

Frequently asked questions

Is Google AI Studio free?

You can start free. As of Nov 2026 new accounts begin on the Free Tier, with access to certain models up to their free tier rate limits. Paid tiers need a linked billing account and a minimum prepay of $5.

Is Google AI Studio the same as the Gemini app?

No. The Gemini app is a general-purpose chatbot. AI Studio is a workspace for testing prompts, changing model settings and building apps that use the Gemini API.

What are system instructions in AI Studio?

They're the instructions in the Run settings panel that tell the model who it's meant to be and how to answer. Google's quickstart turns a model into a Europa alien chatbot with a single line, and you can edit them after a chat has started.

Can people see my API key if I share a Build mode app?

No. AI Studio stores the key as a server-side secret. But their API calls count toward your usage limits, and people you share with can see and fork your code.

What happens when my AI Studio credits run out?

The Gemini API only serves requests with a positive Prepay balance. At zero, every API key in projects linked to that billing account stops, with an HTTP 402 error, until you add credits.