Google AI Studio

Google AI Studio

Google AI Studio is Google's free developer tool for trying Gemini models in the browser, testing prompts and issuing API keys.

Google AI Studio is a developer tool for trying Google's generative AI models, Gemini, in the browser: testing prompts, tuning parameters and issuing API keys. If the Gemini app is the chat product for everyday users, AI Studio is the lab you use before building AI into an app or business system.

What you can do

  • Prototype prompts: change the system instruction, parameters such as temperature, and the output format (for example JSON), and compare results
  • Multimodal input: feed in images, audio, video and PDFs
  • Test tools: grounding with Google Search, code execution and function calling
  • Export code and get an API key: turn a tested prompt into Python or JavaScript and call it directly through the Gemini API
  • Prototype apps: describe the app you want in plain language, get a working web app, and publish it on Google Cloud

Free tier vs paid tier

AI Studio itself is free. The Gemini API also has a free tier that lets you call the main models within request and token limits, while some models, such as video and image generation, are paid only.

The point most often missed is data handling. Google's official pricing page states that content sent on the free tier is used to improve Google's products, and content on the paid tier is not. Run tests that involve customer data or internal documents on the paid tier or on Vertex AI in Google Cloud.

From prototype to production

  1. Write the role and constraints in the system instruction and prepare a few dozen test inputs
  2. Compare models, temperature and maximum output tokens, then settle on a configuration
  3. Export the code and keep prompts under version control so changes can be traced
  4. Move production workloads that need authentication, auditing, regional control or an SLA to Vertex AI

Models are updated often, and the same prompt can start producing different results. Keep an evaluation set and rerun it after every update. For prompt writing, see prompt engineering and system prompt. Our guide to designing an AI PoC covers the planning steps, and we support everything from PoC to production through our AI & DX services.