RAID LogoResponsible AI Development

Lesson 1.1 - Your First API Call

2 min read

Now let’s make your first LiteLLM call!

📓 Work in lesson1-litellm-exercise.ipynb - You can reference lesson1-litellm-solutions.ipynb if you get stuck.

Setup

First, import the necessary libraries:

import litellm 
from dotenv import load_dotenv
import os
 
# Load .env file for non-codespace users
load_dotenv()
OLLAMA_MODEL = f"ollama_chat/{os.environ['OLLAMA_MODEL']}"

What’s happening here:

  • litellm: The main library we’re using
  • dotenv: Loads environment variables from a .env file (useful for local development)
  • os: Used to access environment variables via os.environ
  • OLLAMA_MODEL: Pre-configured local model using an f-string for string formatting

Basic Message Structure

All LLM interactions use a messages list format. Each message is a dictionary with a role and content:

messages = [
    {"role": "user", "content": "What is the capital of Australia?"}
]
 
response = litellm.completion(
  model="gemini/gemini-2.5-flash", 
  messages=messages,
)
 
print(response.choices[0].message.content)

Understanding the response:

  • response.choices[0]: The first (and usually only) completion choice
  • .message.content: The actual text response from the model

Model naming conventions:

  • OpenAI: Just the model name (e.g., "gpt-4")
  • Gemini: "gemini/model-name" (e.g., "gemini/gemini-2.5-flash")
  • Ollama: "ollama_chat/model-name" (e.g., "ollama_chat/llama3")

See the full list of supported providers in the docs.

Switching Models

One of LiteLLM’s superpowers is seamless switching between cloud and local models. Try using the local Ollama model:

response = litellm.completion(
  model=OLLAMA_MODEL,  # Using local model instead of cloud API
  messages=messages
)
 
print(response.choices[0].message.content)

Why use local models?

  • 🆓 Completely free (no API costs)
  • 🔒 Privacy (data never leaves your machine)
  • ⚡ No rate limits
  • 🚀 Great for development and testing

Your Codespace has Ollama pre-installed with a phi4-mini model ready to use. Learn more about Ollama integration.

Key Takeaways

✅ Messages are formatted as [{"role": "...", "content": "..."}]
✅ Use litellm.completion() with model and messages parameters
✅ Switch between any provider by just changing the model name
✅ Access responses via response.choices[0].message.content

What’s Next?

In the next lesson, you’ll learn how to stream responses token-by-token for a better user experience.

Previous: Lesson 1.0 - Introduction to LiteLLM
Next: Lesson 1.2 - Streaming Responses

Graph View

Backlinks