OpenRouter API

Instead of hard-coding one specific model, we'll use the special model ID openrouter/free. It's a router: for each request, OpenRouter automatically picks an available free model that supports the features our request needs (like the tool calling we'll add later). This keeps us on the free tier without having to babysit which models happen to be available on any given day.

You can think of tokens as the currency of LLMs. They're the way that LLMs measure how much text they have to process. Tokens are roughly 4 characters for most models. It's important when working with LLM APIs to understand how many tokens you're using.

1. Create an account on OpenRouter if you don't already have one.

2. Go to the Keys page and click "Create Key". Give it a name like bootdev-ai-agent.

   While creating the key you can set a credit limit on it (for example, $1). This caps how much the key could ever spend, so even if you add credit later you won't be surprised by a big bill. On the pure free tier the key can't spend anything anyway, but it's a good habit to build.

3. Copy the API key and paste it into a new .env file in your project directory. The file should look like this:

   OPENROUTER_API_KEY='your_api_key_here'
   

4. Add the .env file to your .gitignore.

   We never want to commit API keys, passwords, or other sensitive information to Git.

5. Update the main.py file. When the program starts, load the environment variables from the .env file using the dotenv library, and read the API key:

   import os
   from dotenv import load_dotenv
   
   load_dotenv()
   api_key = os.environ.get("OPENROUTER_API_KEY")
   

6. If the environment variable wasn't found - i.e., if api_key is None - raise a RuntimeError with a helpful message.

7. Import the OpenAI class from the openai library and create a client. The key trick is that we point the client's base_url at OpenRouter instead of OpenAI's default:

   from openai import OpenAI
   
   client = OpenAI(
       base_url="https://openrouter.ai/api/v1",
       api_key=api_key,
   )
   

8. Use the client.chat.completions.create() method to get a response from the model. You'll need two named parameters:

   • model: the model ID, openrouter/free
   • messages: a list of message objects. For now, just a single user message. Each message is a dictionary with a role and content. Hard-code the prompt exactly like this:

     messages = [
         {
             "role": "user",
             "content": "Why is Boot.dev such a great place to learn backend development? Use one paragraph maximum.",
         }
     ]
     

9. The method returns a chat completion object. The model's text answer lives at response.choices[0].message.content. Print it to see the model's answer.

If everything is working as intended, you should be able to run your code and see the model's response in your terminal!

Submit the CLI tests.