Gemini API Exercises — Practice Coding

1

Initialize the Google Generative AI client

import { } from "@google/generative-ai";

const genAI = new GoogleGenerativeAI(process.env.GEMINI_API_KEY!);

2

Get a model instance from the client

const model = genAI.({

model: "gemini-1.5-flash"

});

3

Generate text content from a prompt

const result = await model.("Explain closures in JavaScript.");

4

Extract the text string from a Gemini response

const result = await model.generateContent(prompt);

const text = result.response.();

5

Add a system instruction to the model config

const model = genAI.getGenerativeModel({

model: "gemini-1.5-pro",

: "You are a concise coding assistant."

});

6

Start a multi-turn chat session

const chat = model.();

const response = await chat.sendMessage("Hello!");

7

Force structured JSON output via generationConfig

const model = genAI.getGenerativeModel({

model: "gemini-1.5-pro",

generationConfig: { : "application/json" }

});

8

Count tokens before sending a large prompt

const { } = await model.countTokens(largePrompt);

if (totalTokens > 900_000) throw new Error("Too large!");