cloudparts

Google AI Studio

Google’s browser workspace for the Gemini API: try Gemini models with prompts and tools, get API keys and manage billing for the Google Cloud projects behind them, and build apps from prompts that deploy to Cloud Run.

What it is
Large language models
Used for
AI app builders
Type
Commercial product
Made by
Google LLC
License
Proprietary
Facts checked
13 October 2026
Website

What it is

AI Studio is the developer console for the Gemini Developer API. In the Playground you write a chat prompt with system instructions, change model parameters and safety settings, and turn on tools such as structured output, function calling, code execution and Google Search grounding; “Get code” then gives the same request in your programming language. Every API key belongs to a Google Cloud project: a new user gets a default project and key after accepting the terms, and existing Cloud users import their projects. Since 28 May 2026 new keys are authorization keys bound to a service account, and the API rejects unrestricted standard keys. Rate limits apply per project, not per key, in requests and tokens per minute and requests per day, which reset at midnight Pacific time.

New accounts start on a free tier with access to certain models at lower limits. Under the Gemini API terms, AI Studio itself and free API quota are “Unpaid Services”: Google uses the prompts and responses to improve its products, and human reviewers may read them. Linking a billing account and prepaying at least $5 moves a project to the paid tiers, where Google doesn’t use prompts or responses to improve its products. Tiers 1 to 3 raise limits as cumulative spend grows and come with spend-based rate limits ($10, $50 and $200 per 10 minutes) and monthly caps on the billing account ($250, $2,000, then $20,000 to $100,000 or more).

Build mode turns a prompt into an app. The Antigravity Agent, the agent harness from Google’s Antigravity, writes a React front end and a Node.js server, or a Kotlin and Jetpack Compose Android app, installs npm packages, keeps secrets server-side and can set up Firebase Firestore and Authentication. Projects sync both ways with GitHub or download as a ZIP. Publishing creates a Cloud Run service: eligible personal accounts can deploy up to two apps in one region on the Google Cloud Starter Tier without a billing account, and everyone else needs a Cloud project with billing enabled.

Why teams pick it

  • The shortest path from a prompt to Gemini API code: sign in with a Google account, and a project and API key are created for you
  • Tools such as Google Search grounding, code execution, function calling and structured output can be tried in the browser before writing any code
  • Build mode produces full-stack web apps or Android apps with a server runtime, Firebase backend and GitHub sync, and publishes them to Cloud Run
  • Paid tiers keep prompts and responses out of Google’s product improvement, and prepaid credit plus spend caps bound what a project can cost

Watch out for

  • Free use is paid for with data: prompts and responses in AI Studio and on the free API quota can be used to improve Google products and may be read by human reviewers
  • Google’s models only: Gemini, Gemma and Google’s image, video and audio models, not a console for other vendors
  • Models turn over fast: a deprecated model leaves the model picker and API calls to it are routed automatically to its replacement, so output can change without a code change
  • Limits are per project and tighter for preview and experimental models, and the free Starter Tier deploys only two apps, in one region, and excludes Workspace accounts

How it fits with other technologies

Call Gemini with a key from AI Studio · bash

export GEMINI_API_KEY=<key from aistudio.google.com/apikey>

curl -X POST "https://generativelanguage.googleapis.com/v1beta/interactions" \
  -H "x-goog-api-key: $GEMINI_API_KEY" \
  -H 'Content-Type: application/json' \
  -d '{
    "model": "gemini-3.8-flash",
    "system_instruction": "You are a cat. Your name is Neko.",
    "input": "Hello there"
  }'

See it in a real stack

No architecture uses Google AI Studio yet. Drop it into the playground and wire it up yourself.Open playground →
Where it came from2023–2026

Google opened MakerSuite to public preview at Google I/O on 11 May 2023, as a browser tool for prototyping with the PaLM API. On 13 December 2023, with the public release of the Gemini API, it became Google AI Studio: “a free, web-based developer tool” to develop prompts and then get an API key, with free access to Gemini Pro and Gemini Pro Vision at up to 60 requests per minute, and Vertex AI as the next step for companies that needed data control and governance.

It grew from a prompt playground into the Gemini API’s console for keys, projects, usage, logs and billing. In March 2026 Google rebuilt Build mode around the Antigravity coding agent, with full-stack runtimes, databases and multiplayer apps, and Prepay and Postpay billing plans took effect. From 28 May 2026 new API keys became authorization keys tied to service accounts, so leaked keys can be shut off quickly.

Why it’s built this way

Free, in the browser, with a Google account
Google wanted developers building with Gemini at once, with a free quota it called 20 times larger than other free offerings at launch.
“Get code” rather than keeping work in the tool
A prompt becomes an API call in your IDE or a Colab or Android Studio template, and enterprise workloads can move to Vertex AI.
Keys tied to Google Cloud projects
Projects already handle billing, collaborators and permissions, so AI Studio gives a lightweight view over them instead of a separate account system.
Build mode on the Antigravity agent
One agent that keeps context across a whole project and many files lets a prototype become a full-stack app with a server, database and sign-in.

Timeline

  1. 2023

    MakerSuite enters public preview with the PaLM API (11 May)

  2. 2023

    Relaunched as Google AI Studio with the Gemini API (13 December)

  3. 2026

    Full-stack Build mode with the Antigravity coding agent (18 March)

  4. 2026

    Prepay and Postpay billing plans take effect (23 March)

  5. 2026

    New API keys become authorization keys (28 May)

Sources10 · facts checked 13 October 2026

Improve this page on GitHub