Google AI Studio for Beginners: Master the Basics of Google AI Studio

AI is taking the world by storm. Every day, there is a breakthrough that changes the future of computing.

Google is one of the most popular tech companies and has its own Gemini models to integrate and use on various AI applications. To make it easy for developers to integrate AI into their apps, Google launched Google AI Studio in 2023, focusing on developers to make it easy for them to test and integrate Gemini AI models into the apps and software without complex setup.

Google AI Studio has become popular not just among developers but also among casual users. It is used for personal tasks like personal assistants, transforming images, summarising PDFs, being a study assistant, and more.

Since the Google AI Studio app is new, Google has made several improvements to make it more user-friendly and recently revamped the AI Studio interface to make it easy for the beginner.

However, the Google AI Studio was designed with developers and technical users in mind, and it is new. There are fewer tutorials for beginners to get started with the Google AI Studio; here is my beginner-level guidebook to help users learn how to use Google AI Studio, whether you are a beginner or a developer using Google AI Studio for the first time.

If you are looking for a specific feature or use case, use the table of contents below to quickly navigate to the specific section to find what you are looking for.

What you need to know:

What is Google AI Studio

Google AI Studio is a web-based tool that lets you create, test, and customize prompts using various Gemini models. The tool was primarily built for developers to test and customize the prompt to integrate Gemini models into their products.

For developers, Google Gemini Studio offers a prompt sandbox, which allows developers to edit and test prompts using various Gemini models without setting up their back-testing environment.

You can test various Gemini models, and to help you better understand and compare different capabilities of the Gemini models, Google also displays token limits, performance, and pricing between models, helping developers easily differentiate and build prompts with further customized features like variable inputs, format AI responses as structured JSON directly in your prompt, multi-model support, safety setting configuration and more.

Once you have created the model, you can easily test and run the AI responses in the same browser without requiring you to push the code into other apps and set up a separate IDE.

You can easily export the model with a code with the get URL or the Python code snippet to add to your apps with the pricing details.

The Google Gemini AI Studio is available for developers and regular users with a simple Google account. Unlike most developer tools, Google Gemini AI Studio also doesn’t require any credit card to sign up, and it is completely free to use.

Anyone with a simple Gmail account can sign in to the tools on the web and start creating and using the Gemini AI Studio for free.

The tool is also becoming a great option among casual users for creating personalized AI tools for personalized results. Unlike the Gemini app or even the ChatGPT, Google AI Studio allows the creation of custom prompt templates and lets you control the output.

You can save and reuse your prompts on Google AI, which is currently not possible in the Gemini or the ChatGPT. So you don’t have to rewrite the code again by saving your time.

This is especially useful for repetitive tasks like writing and summarising the text. Google Gemini Studio can also read long PDFs or images and extract information, which has limitations in the Gemini and ChatGPT.

Other benefits are customizable outputs, workflow creation, and better control over your prompts and outputs, making it a powerful tool even for casual users who want more control over AI tools.

Another major highlight feature of the AI studio is the Google Stream. You can share the webcam or screen to get real-time guidance from AI.

Getting Started with Google AI Studio

Currently, Google AI Studio is only available on the web.

• Visit https://aistudio.google.com/, sign in with your Google account, and use the Google AI Studio.

Since it is only available on the web, there are no hard cap minimum requirements specifications; if the web browser is working on your browser, you can access the AI Studio on any web browser.

You can also use Gemini AI Studio on mobile. The site is also optimized for mobile use without any issues.

Let’s Understand the Interface

Before going to the how-to-use Google AI Studio tutorial, let’s understand what various icons, buttons, and sections mean on Google AI Studio to give a better idea when navigating.

Home Page

After signing up for the Google account, you will be redirected to the Google AI Studio home page, which is the main home page where you will spend most of your time tuning or creating AI tasks.

We have plenty of options here, from model selection to completely new icons we never encountered if you haven’t used AI tools before. I have divided the entire page into different sections, and you can follow the meaning of the different features in the different sections below.

First, let’s start with the left sidebar, which allows you to switch between chat, stream, explore Gemini integration, view your past conversations, and save the chat to Google Drive for later reference. Here is what each feature means in details

1. Chat: This is the main area where you can chat with the AI. Similar to how you use ChatGPT and Gemini. It is selected by default when you open the Gemini Studio.
2. Stream: This allows real-time interaction with the Gemini model using voice, webcam, or screen sharing. You can use your microphone to talk to the Gemini in real life or use your webcam to share content from the screen. Whether through text, audio, or video input and output, you can talk to the Gemini model and interact with it like a natural human with your natural voice. As a developer, you can use this feature to create applications requiring real-time assistance and more.
3. Starter apps: If you are struggling with ideas, Gemini offers built-in projects such as video analyzer, text analyzer, and Google Maps explorer, where you can start building these apps, customize them, or clone them and use them in your projects by cloning them from the GitHub.
4. History: You can view all your previous chats, prompts, or interactions. You must connect with Google Drive to save and view all your history.
5. Enable Save: You can tap this button to connect your account to Google Drive and store all your past information in a single place.

Google AI Studio – Run Settings Panel Explained

The RUN settings in the Google AI Studio allow you to customize the model you are using. You can further customize your experience. You can select and use different Gemini models from these settings, control how creative and predictable the output will be, and determine whether the AI can execute the IDE, access real-time information, call your function, return structured data, and more.

These settings are more useful for the developer than the casual user configuration of the AI outputs. It comes with several features, and here is what each setting means.

1. Model Selector: The model selection allows you to select which version of the Gemini model to run your tasks with. Currently, it includes the latest Gemini 2.5 Pro to Gemini 1.5 Flash. You can click on the drop-down menu and select the model based on your preference; you can hover the model further to see the rate limits, pricing tokens, and more.
2. Temperature: With the temperature setup, you can control how creative the model output is. By default, it is set to 1 in the middle; however, if you want less, you can use low settings, or if your model involves a lot of creative tasks like coding and maths, you can set the model to high. Remember, the more the number, the time it takes the model to generate the output; however, it can also depend on the model type and the prompt the user generates.
3. Structured Output: Next is structured data, which you can use to enable AI to give answers in structured formats using JSON, bullet points, or tables. When you enable this, you can click on the edit to further structure and customize the structured output. For casual users, it also offers visual editors.
4. Code Execution: This will enable the model to run code execution or be unsuitable for useful coding tasks, maths, and quick calculations.
5. Function Calling: Useful for developers, function calling allows you to call specific function types in the app or while connecting to the API, plugins, and other servers in their apps or services. You can either edit this using the visual and code editor to customize and list specific functions to be called by the app.
6. Grounding with Google Search: Most models these days can offer real-time information. Suppose your model involves verifying facts and current trends and wants the latest and updated information from your web. In that case, you can enable this feature,e, and the model can access the real-time results from the web using Google Search and can answer up to questions by the user.
7. Reset button: Finally, a reset button allows you to reset all the settings to the default.

Advanced Run Settings Explained

In the Run settings, tapping on the advanced settings gives you further customization options, including the ability to configure model safety settings, add stop requests, limit model length, adjust the top-p in other creativity of the model using words, and more; here is what each feature means in detail.

1. Safety Settings: First are the safety settings, which allow you to edit how the model can respond safely to our users. It includes further sliders for harassment, hate, danger, and more.
2. Add Stop Sequence: Add stop sequence allows you to stop at the specific sequences when giving an output. You can use the string or the specific output types to stop at specific places in the output.
3. Output Length (Tokens): Using this setting, you can control how much output length the model should generate in the one response. By default, it is set to 8192 tokens, roughly 6000 words. You can further customize this and increase tokens. You can even go to 1 million tokens based on the model you are using and ensure that more tokens mean more charges.
4. Top P (Nucleus Sampling): Top P is known as the Nucleus sampling, which allows you to adjust the diversity of the model. A low top P value is more focused, and a high top P value means the model will be more diversified and creative when responding to user queries.

Google AI Studio — Prompt Editor Top Bar Explained

The main interface is where you will spend more time. Google has simplified the interface with the latest updates. It also comes with a toolbar with different options, and here is what they mean.

1. System Instructions: The system instructions allow you to set custom instructions for your output. You can describe the output. You can describe the instruction based on the natural language. This option is useful for both developers and casual users.
2. Share: You can share the prompt using this share icon; however, connect to your Google Drive account to get the shareable link.
3. Get code: For developers, after you finish the prompt and get the desired output, you can quickly copy the code of the prompt and add it to your service.
4. Clear: You can reset the current chat history with a single click.
   Compare outputs: If you are a developer, you can compare different outputs between models.
5. Vertical three-dot menu: Clicking on the three-dot menu gives you more options.
6. Attach media (Plus icon): While you are prompting, you can attach media like links, documents, or an image when you are prompting for the app. Building and Testing Prompts

Writing effective prompts

Once you get familiar with all the icons and features in the Google AI Studio, it’s time to go straight into building prompts. As a developer, you can play with all the tools, settings, and advanced features to customize the AI output based on your requirements.

However, if you are a casual user, you can play with high-end Gemini models to get what you want. We can further customize the model or put output to max all that time, but the best part? Google Gemini AI Studio is free to use.

Considering that Google Studio currently has many people for general AI tasks. As of writing this guide (April 2025), the current rate limits include 1 million tokens per chat and 1500 chats per day, more than enough for casual users or developers to find models for better output. Once all this is in mind, let’s go to the prompt builder and build our prompts.

I am not going into the details for developers; You can refer to this documentation for more technical details. This guide will focus mostly on casual users using Google AI Studio.

How to Write a Prompt

Whether you are a developer or a casual user, writing effective prompts will always be helpful to get better output for the prompt or other tasks. If you are a developer, it allows you to fine-tune the model based on your requirements. If you are a developer already, you need to follow a specific format to better refine the output for your specified product.

However, if you are a casual user, you can directly chat with the prompt using natural language, and Gemini is smart enough to understand. However, suppose you still want to fine-tune your model. In that case, you can follow the principles or use other ChatGPT or other Gemini apps to fine-tune the model further or convert existing natural language into AI-generated prompts.

• Use the structured output to get output in specific formats such as structured output.
• To enhance the model, you can further iterate the input with news and data.

You can still fine-tune the model with further instructions if your output isn’t what you expected. If your output feels off and random and sometimes inconsistent, you can break the model into you can break prompt into multiple steps and start with a smaller model and follow the clear, prompt engineering techniques like being clear and specifying the exact output format you want to enhance your model further. You can provide examples to give more context to the model.

You can adjust the temperature and tokens to customize your output further and specify system instructions for the specific output. You can also add breakpoints if the output is too long.

• Before writing a prompt, you can customize the Run settings and add custom instructions for pertained output.
• Click on the chat tab in the left sidebar, and then in the input bar, enter the prompt and click on the run.
  
• You can select Google Gemini 2.5 Pro, the latest best model from Google, for better results.
  

Google Stream Real Life

Google Stream Real-Time is another popular preview for developers and casual users to interact with Google Gemini models in real-time from your microphone camera using your device’s Webcam.

If you want to build a product that wants to interact with real-time, such as from the camera data, you can use these models to try your AI and create and implement the AI to your device; these are also great options for casual users to use Google Stream Real-time.

For example, the most popular use case is for education. Many people have been using Google Stream to help solve problems by sharing the screen, including solving coding and maths problems in real-time. This is one of the most useful cases of Google Stream in real-time. You can also use the camera, WebCam, and voice to interact with the model in real-time.

Like the prompts, you can customize Google Stream run settings with further control, such as voice on. Find the model with additional settings. See more details below on how they work and how to use them as a developer or casual user interacting with Google’s real-time streaming.

Google AI Studio Stream Mode Interface Explained

The Google Stream interface is very similar to the prompt with the main interface and run settings in the sidebar. First, let’s understand the complex run settings. Here is what each feature means in the run settings.

1. Main interface
2. Custom instructions: Similar to main prompt instructions, you can set custom instructions to model and personalize your output.
3. Get code: Developers can instantly get the code to add to their product after testing it.
4. No settings
5. Gemini Model Selector: You can switch between the AI models here. Currently, Flash is the only available model, and it is set to default.
6. Output: With this option, you can select the output format, whether the Gemini model should respond to you with both text and audio or only with the text-only model.
7. Voice: Next, you can select the voice if you want a voice output. Currently, there are more than six voices available.
8. Language: The model output can be supported in plenty of languages. English is the default. However, you can tap in the drop and change the model to any language you want.
9. Media resolution: This setting allows you to control how much image content the model interprets. The more the number, the more the model directly understands the image. By default, it is set to 258 tokens per image; you can also scale down to 66 tokens per image.
10. Turn coverage: The track and display view lets the model track and display with the user speaking and interacting. This is especially useful if the model interacts with multiple persons simultaneously, like video calls.
11. Session content: Since it is a continuous session, the session content feature allows you to set the maximum limit of tokens, whether they are words the AI can remember in one go. Here, you can adjust the max-content size and target content size. This is especially useful for developers building apps with long interactions, like coding assistants.
12. Tools: Similar to tools in the test prompt, the stream also offers tools for code execution, function calling, automatic function response, grounding with Google Search, and more.

Next is a chat interface where you primarily interact with Gemini models, and you can talk. Use a Webcam or share your screen to interact with the Gemini model in real-time and get real-time out from the model.

Plenty of examples of casual users using Google Stream in real life, such as people using it to learn code and guide software tutorials.

How to use Google Stream

• Go to the stream tab in the Google Gemini Studio
  
• Select the model you want to use: Talk, Webcam, and share screen.
• If you select talk, allow micro-phone permission and start interacting with the model. If you select webcam, allow web and start using the camera; if you select the Share screen, select the screen you want to share and share the screen.
  
• Once you start the session, You can control the microphone and video and end the session by clicking on the close button in the chat window.

Google AI Studio Starter Packs

If you are a developer and just getting started, you can explore Google AI Studio’s built-in starter apps. There are only three apps: spatial understanding, video analyzer, and map explorer. If you are a beginner and new to development, start developing with pre-built apps. I expect Google to add more apps built on AI Gemini starter apps.

Casual users can also use these apps to upload images or videos and explore the maps. You can visit the starter apps from the sidebar and get access to the three initial starter apps.

Things you can try out: Google AI Studio

1. Create your personal AI Assistant

Google AI Studio can assist you with complex tasks in real-time. With the Google Stream features, you can ask AI what is on your screen, interact with AI through two-way communication, and get help from it.

One of the best use cases on this is using this for learning; whether you are a student solving a math problem or a developer stuck at coding, you can share your screen with Google Assistant and then get real-time guidance from it.

The best part is that it is accurate and helped me navigate the complete Google Cloud API setup. It can also help you guide yourself through daily life scenarios like finding the specific settings, navigating the complex interface, and more.

2. Personalised AI Chatbot

You can also use AI Studio as the personalized chatbot; unlike the regular Gemini app or even the ChatGPT, Google AI Studio, you personalize the output results with custom settings and instructions.

3. Use it for your Software tutorial

You don’t need this guide to get started with the AI Studio; you can ask AI to guide you through the steps. You can use Gemini stream screen share content for detailed step-by-step tutorials to quickly guide you through the software tutorial, such as getting started with the Google AI studio.

Those are the three main cases you can use Google AI Studio. Since it is an AI chatbot, you can use it for various daily tasks such as an AI blog post summarise, weekly plan generator, meal and grocery list tracker, calorie tracker using Gemini AI starter apps, resume build engine, and much more.

Google AI Studio for Everyone

Developing and integrating AI into the apps and making them personalized requires a complex prompting and API setup. Also, AI models, such as large language models, are powerful but complex to set up for the app for personalized results.

This often requires fine-tuning the models, making them content-aware of the situation, and offering customized results. This is where AI Studio can be useful for developers using Gemini models to fine-tune models, create and test various types of prompts, tune models to improve their performance on particular tasks, and easily integrate the model using Gemini AI.

Google AI Studio is not just for developers; it is also very useful for regular users. For a visual guide, you can watch this video from Tina, where she walks you through a complete tutorial.

FAQs about Google AI Studio

1. How is Google AI Studio different from the Gemini app I already have on my phone?

For regular users, you can use Google Gemini AI Studio for custom instruction and the very useful Google Stream, which lets you interact with the model in real-time; however, for developers, this is only the tool for creating fine-tuned AI models with custom instructions, file upload exporting API and also save them in terms of projects.

2. Can I share the prompts I create in AI Studio with friends or teammates?

Yes, you can share the prompts you created in the AI studio with friends or teammates, but first, you need to connect your account to Google Drive.

3. Can I connect Google AI Studio to Google Sheets or Docs to automate tasks?

Yes, you can, but as of now, Google AI Studio does not directly support third-party integration like docs. You can use alternative methods like Google API scripts, function calling, or even third-party tools to connect with external apps.

Source link