Building your first LLM API call in Python (step by step)

A developer published a step-by-step guide for making a first LLM API call in Python, walking through setting up a virtual environment, installing the Anthropic SDK and python-dotenv, and calling Claude via the messages.create endpoint. The tutorial script loads an API key from a .env file, creates an Anthropic client, sends a prompt to the claude-haiku-4-5 model with a 1024-token cap, and prints the response while logging token usage.

Before going any further with this build, a quick note. This build assumes you have read aie 1.0. If you have not, start there https://heymeraki.substack.com/p/aie 10-introduction-to-ai-engineering?r=8ei6qe&utm campaign=post&utm medium=web , it’ll make this make more sense. In the previous lesson, we established that an LLM, Large Language Model, is a prediction machine. It has no memory i.e. it is stateless. And as a result, it depends on the context you provide. In this build, we are going to talk to one in code. We’ll call what we build First Contact. What we are building A Python script that makes an API call to Anthropic, prints what it gets back and logs how many tokens the unit a context window is measured in it used. It is intended to be a simple build so you see how LLM calls work. What you need - A terminal - Python installed on your machine - An Anthropic API key platform.claude.com https://platform.claude.com/ , you will need to add a small credit balance as the free tier does not cover API access - A code editor VS Code is fine if you do not have a preference Setting up Create a folder for the project and set up a virtual environment. A virtual environment keeps your project's dependencies isolated, you can think of it as a container for everything this project needs, separate from anything else on your machine. mkdir ai-engineering cd ai-engineering python3 -m venv venv source venv/bin/activate You will know it worked when you see venv at the start of your terminal line. Install the two libraries you need: pip install anthropic python-dotenv - anthropic is the official Python library for talking to Claude. - python-dotenv reads your API key from a file so you never have to hardcode it in your script. Create a .env file and add your key: ANTHROPIC API KEY=your key here Create the script file: mkdir 01 touch 01/first contact.py Open 01/first contact.py in your editor. This is where you will build the script, piece by piece. Step 1: load your API key The first thing the script needs to do is read your API key from the .env file. Without it, every call you make to Claude will fail. python from dotenv import load dotenv import os load dotenv - load dotenv reads your .env file and makes everything inside it available to the script. - import os gives the script access to environment variables, the place where your API key lives after load dotenv runs. After this line runs, your API key exists in the environment and your code can access it. Step 2: create the client Before you can send a message to Claude, you need to create a connection to Anthropic’s API. You can think of it as opening a phone line, you set it up, and every call you make goes through it. In your code, that connection is something called a client. It holds your credentials, knows where to send your requests and it gives you a way to interact with the API without dealing with any of the complications under the hood. python from anthropic import Anthropic client = Anthropic api key=os.getenv "ANTHROPIC API KEY" - os.getenv "ANTHROPIC API KEY" reads the key from the environment. - Anthropic api key=... creates the client using that key. Step 3: make the API call This is the call to Claude. You are sending a message and asking for a response. response = client.messages.create model="claude-haiku-4-5", max tokens=1024, messages= {"role": "user", "content": "What is a large language model? Answer in two sentences."} A few things worth noting here: - model specifies which Claude model to use. - max tokens sets a ceiling on how long the response can be. - And messages , that list with the role and content, is the context window in its simplest form. This is what you send to the model and it is all it can see. Step 4: read the response The response comes back as an object. You need to pull out the parts that matter, in our case this is the text and the token counts. text = response.content 0 .text input tokens = response.usage.input tokens output tokens = response.usage.output tokens - response.content 0 .text is the model’s response as a string. - response.usage gives you the token counts, how many tokens you sent in and how many the model generated back. Step 5: print everything print "Response:" print text print print f"Input tokens: {input tokens}" print f"Output tokens: {output tokens}" print f"Total tokens: {input tokens + output tokens}" Your final code should look like this: python from dotenv import load dotenv import os from anthropic import Anthropic load dotenv client = Anthropic api key=os.getenv "ANTHROPIC API KEY" response = client.messages.create model="claude-haiku-4-5", max tokens=1024, messages= {"role": "user", "content": "What is a large language model? Answer in two sentences."} text = response.content 0 .text input tokens = response.usage.input tokens output tokens = response.usage.output tokens print "Response:" print text print print f"Input tokens: {input tokens}" print f"Output tokens: {output tokens}" print f"Total tokens: {input tokens + output tokens}" Run it python3 01/first contact.py You should see something like this: Response: A large language model LLM is an AI system trained on vast amounts of text data to understand and generate human language. It uses deep learning to predict and produce text by identifying patterns in the data it learned from. Input tokens: 19 Output tokens: 51 Total tokens: 70 What you are seeing The response is the prediction machine predicting, like we said in the last lesson, the most likely continuation of your input. Not based on any fact that it knows, it predicted what a two-sentence response to that question should look like based on its training data. The token counts show what the context window looks like. You sent 19 in my case tokens. The model responded with 51. In total, the entire exchange used 70 tokens. That number is what adds up against the context window limit with every call that you make. Now that you have made your first contact, in the next lesson, we’ll explore everything that went into the call we have just made. We’ll learn what happens when you hit send and learn how to start controlling it. If you have any questions about this build, let me know