Module 2 (Gemini), Lesson 1: Working with Google Gemini

Welcome to Module 2 (Gemini)

You built your foundation with OpenAI in Module 1 and expanded to Anthropic. Now you will add Google Gemini to your toolkit so you can pick the right model for each job and avoid vendor lock-in.

Why Gemini?

• Strong speed/price balance with the gemini-3-flash family
• Native JSON and function calling via responseMimeType, schemas, and tools
• Simple streaming helpers with generateContentStream
• Google ecosystem fit if you already use Firebase, Vertex, or Workspace

What You Will Learn

• ✅ Gemini client setup with @google/genai
• ✅ Prompt patterns that mirror your OpenAI and Anthropic lessons
• ✅ How Gemini structures contents, candidates, and usageMetadata
• ✅ Streaming, structured output, and tool calling the Gemini way

Code Folder for This Module

All lesson code lives in src/gemini/ in the playground repo:

src/
├── gemini/
│   ├── basic-prompt.ts
│   ├── basic-prompt-with-system.ts
│   ├── basic-prompt-with-temperature.ts
│   ├── extended-prompt.ts
│   ├── stream-prompt.ts
│   ├── structured-output-prompt.ts
│   └── tools-prompt.ts
├── anthropic/
└── openai/

Navigation

• Next: Lesson 2: Gemini Prompts Overview
• Module Index: AI SDK Essentials

Ready? In the next lesson you will see the Gemini-specific prompt map for the rest of this module.