Getting Started with the Gemini API: A Practical Tutorial

#ai #python #api #tutorial

Getting Started with the Gemini API

The Gemini API lets you harness powerful language models for a variety of tasks—from generating text to analyzing data. This tutorial will walk you through setting up your environment, making your first request, and interpreting the response, all with a short, runnable Python example.

1. Prerequisites

Python 3.8+ installed on your machine.
requests library (pip install requests).
A Gemini API key from the Gemini console (you’ll need to create an account and enable the API).

2. Set Up the Request

Below is a minimal Python script that sends a prompt to the gemini-3.6-flash:free model and prints the generated text.

import requests
import json

API_KEY = "YOUR_API_KEY_HERE"
MODEL = "gemini-3.6-flash:free"
URL = f"https://api.unorouter.com/v1/chat/completions"

headers = {
    "Authorization": f"Bearer {API_KEY}",
    "Content-Type": "application/json"
}

payload = {
    "model": MODEL,
    "messages": [{"role": "user", "content": "Explain in two sentences what GPT-4 is."}],
    "stream": False
}

response = requests.post(URL, headers=headers, json=payload)
data = response.json()

if response.ok:
    # Extract the assistant's reply
    assistant_reply = data["choices"][0]["message"]["content"]
    print(assistant_reply)
else:
    print(f"Error: {data.get('error', 'Unknown error')}")

How It Works

Headers – We authenticate with the API key and set the content type to JSON.
Payload – The model field selects the free Gemini model. The messages array contains a single user message; you can adjust the prompt as needed.
Response – On a successful request (200 OK), the generated text lives under choices[0].message.content.

3. Running the Script

Save the code to a file, e.g., gemini_demo.py, replace YOUR_API_KEY_HERE with your actual key, and run:

python gemini_demo.py

You should see a concise explanation of GPT‑4 printed to the console.

4. Tips & Best Practices

Rate Limits – The free tier includes a generous number of requests, but monitor your usage to avoid hitting limits.
Prompt Engineering – Clear, specific prompts yield more predictable results. Experiment with temperature settings if the API supports them.
Error Handling – Always check the response status and handle potential errors (e.g., network issues, invalid keys).

5. Next Steps

Explore other models (e.g., gemini-2.0-flash-exp) for different capabilities.
Integrate the API into larger applications using the official SDKs or REST calls.
Build simple tools like text summarizers, code assistants, or chatbots.

With just a few lines of code, you’ve unlocked the Gemini API’s potential. Happy coding!

DEV Community