Google AI Studio Cheat Sheet: Features, Pricing, and More

For years, building software meant setting up local environments, downloading SDKs, configuring dependencies, debugging installations, and spending hours just to write usable code. Google now wants to collapse much of that process into a browser tab.

With the rapid expansion of Google AI Studio, the company is no longer treating AI as just a chatbot layer or productivity assistant. Instead, Google is turning AI Studio into something far larger: a browser-based operating system for AI-assisted software creation.

In 2026, AI Studio sits at the intersection of several trends happening simultaneously across the technology industry:

• Vibe coding
• Multimodal AI generation
• AI-powered prototyping
• No-code development
• Agentic software workflows.

The platform now allows users to generate apps, test Gemini models, create images and videos, build Android applications, deploy cloud projects, export code to GitHub, and connect directly into Google’s wider ecosystem, all without leaving the browser.

The result is one of the most aggressive attempts yet to reduce the friction between an idea and a functioning application. But AI Studio is also messy in places, inconsistent in others, and still clearly evolving.

This cheat sheet takes a deep look at what AI Studio actually is in 2026, what works exceptionally well, what still feels unfinished, and why the platform matters far beyond simple AI experimentation.

What Google AI Studio actually is

At its core, Google AI Studio is a browser-based AI development workspace built around the Gemini family of models. The simplest way to understand it is this:

• ChatGPT is primarily designed to help users interact with AI.
• AI Studio is designed to help users build things with AI.

That distinction matters.

Instead of focusing mainly on conversations, AI Studio behaves more like a lightweight development platform. Users can test models, generate code, prototype interfaces, create media, connect APIs, and increasingly deploy complete applications.

Google has effectively combined:

• AI chat
• Prompt engineering
• Media generation
• App prototyping
• Code export
• Cloud deployment
• Android development inside a single interface.

The platform’s strongest feature is not necessarily any one individual tool. It’s the fact that Google has centralized nearly all of its generative AI workflows within a single browser environment.

How to access it

Access the platform at aistudio.google.com or ai.dev. Sign in with any Google account, accept the terms of service, and you are on the platform immediately.

The interface: Surprisingly dense, occasionally overwhelming

The first thing many users notice is that AI Studio does not behave like a polished consumer product. Unlike OpenAI ChatGPT or Anthropic Claude, AI Studio feels closer to a developer console.

There are multiple workspaces:

Playground

The experimentation hub. Test Google’s AI models in real time. Switch between the Gemini tab for chatbots and agents, or the Image tab for generating visuals. This is where you explore and refine prompts before moving them into production.

Build

The app construction zone. Describe what you want to make in plain language, and Gemini will generate working code, with a live preview that updates as you iterate. This is the vibe coding tab where apps come to life.

Dashboard

Your management center. Oversee all your projects, create and manage API keys, monitor your usage, and track billing if you have enabled it. Think of it as your control tower.

Documentation

In-platform guides covering what each model can do, how to structure prompts, and how to integrate the API into external projects. A useful quick-reference without leaving the tool.

Run settings; the control room inside Playground

Inside Playground, a panel called Run Settings gives you fine-grained control over model behavior. The key controls include:

• Model selection: Switch between Gemini variants at any point.
• Temperature: A slider controlling how creative or random the model behaves. Low temperature gives more predictable, factual responses; high temperature produces more varied, imaginative output.
• Safety settings: Adjust thresholds for harmful content across categories such as harassment, hate speech, sexually explicit content, and dangerous content.
• Stop sequences: Define specific strings that, when generated, will cause the model to stop producing further output.
• Aspect ratio: Available when generating images, letting you set specific width-to-height dimensions.
• System instructions: Define the model’s persona, tone, and behavioral rules before any user message is sent.

AI models available

Google AI Studio gives you access to a range of models, each optimized for different tasks. Here is the current lineup as of May 2026.

Text and reasoning models

• Gemini 3.5 Pro (Maximum intelligence): The most capable model for complex reasoning, long-context analysis, and tasks that demand depth over speed. Best reserved for genuinely difficult tasks where accuracy matters most.
• Gemini 3.5 Flash (Balanced speed and power): The workhorse model faster than Gemini 2.5 Pro and the default choice for most tasks. Excellent for everyday coding, content generation, and chatbot work.
• Gemini 3 Flash Preview (Speed-first lightweight): The fastest and cheapest model. Ideal for simple, high-volume tasks where speed matters more than depth.

Image generation models

• Nano Banana (Gemini Flash Image Preview): Conversational image creation with editing capabilities. Generate an image, then request modifications through natural language prompts. Strong for iterative creative work.
• Imagen 4: Photorealistic image generation with precise technical controls for aspect ratio, resolution, and batch creation. Multiple results per prompt let you compare outputs.
• Nano Banana Pro: The upgraded version of Nano Banana with enhanced capabilities. Handles more complex image editing and generation requests.

Video, audio, and music models

• Veo 3.1: Text-to-video and image-to-video generation. Creates 8-second clips in multiple aspect ratios with customizable frame rates. Capped at 4K (3840×2160) resolution. The free tier allows 10 test videos.
• Voice/TTS: Natural voice text-to-speech synthesis supporting multiple speakers. Useful for creating podcast-style audio, language learning content, and voice interface prototypes.
• Lyria (Music): Real-time music generation based on text prompts. Specify genre, mood, instruments, BPM, and musical scales. An interactive player updates instantly as you modify parameters.

Context window: All Gemini 3 models support a 1-million-token context window, large enough to analyze an entire codebase, a lengthy research paper, or hundreds of pages of documents in a single session.

Key features — deep dive

Chat and message editing

The Chat tab is the foundation of the platform. Beyond basic question-and-answer, it supports several advanced workflows:

• Message editing: Click any message in a conversation to modify it, then rerun the exchange from that point. Saves significant time when fine-tuning prompts.
• Branching conversations: Create alternate conversation paths without losing the original thread. Ideal for A/B testing different approaches or prompting strategies.
• Compare mode: Run the same prompt across multiple Gemini models simultaneously and watch responses stream in real time to identify the best fit for a task.
• Google Search grounding: Enable real-time web search to give the model access to current information beyond its training data.
• Thinking mode: Activate step-by-step reasoning for complex multi-part problems that benefit from slower, more deliberate computation.
• URL context integration: Feed live URLs directly into a prompt so the model can analyze and reference them.

Code export

One of the platform’s most practical features is how easy it makes exporting working code. In Playground mode, clicking Get Code converts the current prompt configuration into your chosen programming language, such as Python, JavaScript, REST API, and more.

In Build mode, the export options expand considerably. You can copy and paste code from the Code tab, push directly to a GitHub repository, or download the entire app as a ZIP file. The exported code is described by developers as clean, well-structured, and free of vendor lock-in.

System instructions

System instructions are among the most powerful and underused features.

They sit above the conversation and define the model’s personality, role, constraints, and tone before any user message arrives. Developers use them to create consistent AI behaviors, a formal support assistant for one project, a playful language tutor for another. The instructions persist across the entire conversation without needing to be repeated.

Screen streaming

A newer addition to the platform, Screen Streaming lets you share your screen with the model to receive real-time AI guidance with full visual context. This is particularly useful for debugging interface issues, walking through presentations, or getting contextual feedback on what you are looking at.

Google Colab integration

Generated code can be sent directly to Google Colab for immediate execution in a notebook environment. Researchers and data analysts use this flow to move from prompt-tested logic to running code in a single click.

Google Workspace integration

Announced at Google I/O 2026, AI Studio can now build apps that connect directly to Google Sheets, Drive, and Docs without switching tools. Teams already embedded in the Google ecosystem can build internal tools that read and write to their existing data sources without any API setup.

Annotation tool

A new visual feature that lets you draw directly on the app preview window to mark up, tweak, and annotate components. You can also use the annotation layer to trigger the generation of new visuals within the same workspace. This brings AI Studio closer to a design-and-build tool rather than just a code generator.

Build Android apps without writing a single line of code

The most significant recent update to Google AI Studio, announced at Google I/O 2026 this week, is the ability to build native Android applications directly from a text prompt inside a web browser. This is not a web app wrapped to look like a mobile app; it is real, production-quality native code.

What this changes: Traditional Android development required downloading massive SDKs, configuring local environments, and deep knowledge of Java or Kotlin. Google has eliminated that entire setup process. Anyone can go from a prompt to a testable app in minutes.

What it generates

AI Studio generates apps using Kotlin and Jetpack Compose, the modern, Google-recommended stack for native Android development. These apps use the actual Android SDK, which means they can access:

• GPS location data 
• Bluetooth connections 
• Accelerometer/gyroscope 
• Onboard camera 
• Background tasks 
• Offline operations

The browser-based Android emulator

Embedded directly into the AI Studio browser window is a cloud-hosted Android Emulator. As the model generates or updates code, you can immediately click, swipe, and test the interface in real time.

Deployment workflow

• Prompt and preview: Describe your app idea. The model generates Kotlin code and shows a live preview inside the emulator.
• Iterate: Use the annotation tool or chat interface to request changes. The emulator updates in real time.
• Install on device: Connect a physical Android phone via USB and install the app directly using integrated debugger tools.
• Publish to Google Play: Connect your Google Play Developer account to automatically bundle and publish the app to an internal testing track in minutes.
• Export for advanced work: Download as a ZIP or push to GitHub to continue development in Android Studio or with a professional team.

Current capabilities

The platform currently supports single and multi-screen utilities, basic social frameworks, and custom Gemini API integrations. Cloud database tools are scheduled to launch in the coming update.

Also coming: Google is launching a mobile app for AI Studio, available for pre-registration at Google I/O 2026, so you can build and iterate on apps directly from your phone.

Media generation: images, video, audio, and music

Google AI Studio is increasingly a multimodal production environment, not just a text tool. The Generate Media section consolidates image, video, audio, and music creation in a single interface.

Image generation

Two models serve different workflows. Nano Banana (available free) is built for conversational image creation. You generate an image, then modify it through follow-up natural language prompts rather than re-entering a full new prompt each time. Imagen 4 (paid tier) is designed for precision and scale, offering technical controls for aspect ratio, output resolution, and batch generation.

Video generation

The Veo 3.1 model creates 8-second videos from text prompts or input images. Frame rates and output resolution are customizable, and the model supports multiple aspect ratios. Simple, clear prompts tend to produce the best results; overly complex scenarios can confuse the model.

Audio and voice

Multi-speaker audio generation is one of the platform’s most distinctive features. You can create podcast-style conversations between multiple AI voices: define each speaker’s voice, write dialogue in script format, and generate a full audio track. Language learning applications are a natural fit; you can generate native-sounding conversations in target languages for practice purposes.

Music with Lyria

The Lyria model generates real-time music from text descriptions. You can specify genre, mood, instrumentation, BPM, and even musical scales. An interactive player updates as you adjust parameters, letting you fine-tune the output before exporting. Results vary, but the tool can produce credible backing tracks and ambient music for prototyping.

Pricing: free tier vs. paid tier

Google AI Studio’s pricing model is one of its biggest selling points. The free tier is generous enough that most individual developers and small businesses will never hit its limits during normal usage.

• • Free tier ($0/month)
  • Google AI Plus ($7.99 / month) 2x higher usage access than Free
  • Google AI Pro ($19.99/month) 4x higher usage access than Free

• Google AI Ultra 

• • $99.99/month: 5x higher usage limits vs. AI Pro
  • $199.99/month: 20x higher usage limits vs. AI Pro
• Pay-as-you-go (API) usage-based

Privacy note: On the free tier, Google collects and may use your prompts, uploaded files, and generated content to improve its models. Human reviewers may occasionally inspect inputs. If your work involves sensitive business data, enable billing to activate privacy protections. Paid plans provide you with the controls to opt out, whereas enterprise accounts have it disabled by default

Google AI Studio vs. the competition

When to pick each platform

Choose AI Studio when…

You need free access for prototyping, massive context windows, built-in image/video/audio generation, native Android app output, or cost-effective API pricing. Especially strong if you are already in the Google ecosystem.

Choose ChatGPT when…

You need cross-session memory, a large plugin ecosystem, or a consumer-friendly interface for mixed everyday tasks. Better for individuals who use AI for personal productivity rather than app development.

Choose Claude when…

Coding quality is the top priority, privacy is non-negotiable, or you are doing iterative long-form writing and enterprise compliance work. Has the most consistent no-training-on-user-data policies by default.

Choose Lovable/Bolt when…

You need full-stack web deployment in one click, built-in database support (Supabase), or want to build complete web apps with complex backends rather than AI-focused prototypes.

Real-world use cases

Here is how businesses and individuals are actually using Google AI Studio across industries.

• Android app prototyping: Startups and indie developers are going from idea to testable Android app in under an hour, then deploying directly to Google Play’s internal testing track.
• Customer support chatbots: E-commerce companies build AI systems trained on documentation and FAQs to handle first-line support. 
• Content and marketing: Marketing agencies use reusable prompt templates to standardize content generation across multiple clients, consistent tone, and faster output.
• Research and document analysis: Researchers upload entire papers or datasets and use a 1M-token context to extract insights from unstructured data without manual summarization.
• Code review and debugging: Development teams upload codebases for review, documentation generation, and bug detection before shipping. The context window can handle thousands of lines at once.
• Language learning tools: Voice capabilities allow creation of native-sounding multi-speaker language practice conversations, something that would otherwise require multiple external APIs.
• Visual search and retail: Retail apps use Agentic Vision (AI-assisted image analysis) to identify products in customer photos and automatically surface relevant information.
• Internal business tools: Teams build budget trackers, email drafters, SEO analyzers, and alerts dashboards without external developers. 

Limitations you need to know

Google AI Studio is genuinely impressive, but no platform is without constraints. Understanding where it falls short will save you from unpleasant surprises mid-project.

Locked to Google’s models

You cannot bring in OpenAI’s GPT models, Anthropic’s Claude, or open-source alternatives like Llama or Mistral. Every task runs through Google’s Gemini ecosystem. This is fine for most use cases, but teams that need multi-model workflows or model redundancy will need to look elsewhere.

Rate limits (and recent cuts)

The Dec. 2025 rate limit reductions significantly tightened free tier access. Gemini 3 Pro now allows only 50 requests per day, a meaningful drop from previous limits. Power users testing intensively will hit these walls quickly.

Data privacy on the free tier

Google may use your inputs, prompts, and uploaded files to improve its models on the free plan. If your project involves proprietary business information, client data, or anything commercially sensitive, opt out before using the platform for serious work.

Deployment complexity

While AI Studio makes building easy, hosting production apps outside of Google Cloud requires manual work: setting up API key management, building authentication layers, configuring database persistence, and hardening the generated code for security. The generated apps work well as prototypes, but are not production-ready out of the box.

Platform stability

In early 2026, the platform experienced some stability issues: legacy chat sessions failing to migrate to the new reasoning engine, internal errors with large file uploads, and occasional inconsistencies between AI Studio behavior and Vertex AI behavior. These appear to be stabilizing but are worth noting for teams that depend on consistent uptime.

Pro tips and best practices

Experienced developers and agencies that use the platform daily have identified patterns that distinguish efficient workflows from frustrating ones.

• Spend an hour in Chat mode before building anything: Getting a feel for how the model responds to your instructions before moving into code generation prevents hours of debugging misaligned behavior later.
• Build a personal prompt library: Save prompts that work. Agencies using reusable templates across multiple clients report significant time savings compared to starting from scratch on each project.
• Use Flash first, upgrade to Pro only when needed: Flash handles most tasks reliably and is far more cost-effective. Reserve Pro for genuinely complex reasoning tasks that require maximum capability.
• Export code early and often: Don’t wait until the prompt is perfect. Export early versions, test them in a real environment, and iterate on both the prompt and the implementation in parallel.
• Monitor your daily quotas proactively: Free tier limits are generous for normal use, but can be consumed quickly during intensive testing. Consider enabling billing before hitting limits to avoid disruptions.
• Use the 1M token context aggressively: Upload entire codebases, lengthy contracts, full research papers, or complete meeting transcripts. Gemini’s context window is one of its biggest technical advantages; fully leveraging it is where real productivity gains occur.
• For Android apps test on a real device early: The browser emulator is excellent for quick iteration, but install on a physical phone via USB before too much time passes. Hardware-specific behavior (especially GPS, Bluetooth, and camera) can differ from emulation.

Final word

Google AI Studio is not a product that needs to be oversold.

The facts speak plainly: a free, no-credit-card-required platform that can produce native Android apps from text prompts, generate images and video, create multi-speaker audio in any language, and export clean, working code to GitHub or Google Cloud is a genuinely remarkable piece of infrastructure made freely available to anyone.

The barrier to entry for building real, functional AI-powered products has never been lower. Google AI Studio is one of the primary reasons for that.

Editor’s note: This article originally appeared on our sister publication, eWeek.