[Local AI with Ollama] Using Python Functions with the Ollama Library

The official Ollama Python library provides a high-level, Pythonic way to work with local language models. It abstracts away raw HTTP requests and makes model management, chatting, and customization much easier and more readable. In this guide, you'll learn how to interact with Ollama models using Python functions—covering everything from listing models to chatting, streaming, showing model info, and managing custom models.

Prerequisites

- Ollama should be installed and running locally

- Make sure the Ollama service is running on your machine.

- Set up your Python environment

- Create and activate a virtual environment:

python3 -m venv venv
source venv/bin/activate

- Install the Ollama Python library

pip install ollama

Listing Available Models

You can list all models currently available in your Ollama instance:

# list_models.py
import ollama

models = ollama.list()
print(models)

How to run:

python list_models.py

Chatting with Models

You can have multi-turn conversations with models using the chat function. This is ideal for building chatbots or assistants.

# chat_tiger.py
import ollama

response = ollama.chat(
    model="llama3.2",  # Replace with your model name
    messages=[
        {"role": "user", "content": "Tell me about tigers."}
    ]
)
print(response['message']['content'])

How to run:

python chat_tiger.py

Streaming Responses

For long or real-time outputs, you can stream model responses as they are generated. This is useful for chat UIs or applications that need to display content as it arrives.

# stream_tiger.py
import ollama

for chunk in ollama.chat(
    model="llama3.2",
    messages=[{"role": "user", "content": "Tell me about tigers."}],
    stream=True
):
    print(chunk['message']['content'], end="", flush=True)

How to run:

python stream_tiger.py

Using the show Function

You can inspect detailed information about any model (metadata, creation time, config, etc.):

# show_model.py
import ollama

info = ollama.show("llama3.2")
print(info)

How to run:

python show_model.py

Generating Text (Single Prompt)

If you want to generate text from a single prompt (not a chat), you can use the generate function:

# generate_ocean.py
import ollama

result = ollama.generate(
    model="llama3.2",
    prompt="Which is the largest ocean in the world?"
)
print(result['response'])

How to run:

python generate_ocean.py

Creating and Managing Custom Models

You can create, use, and delete custom models directly from Python with the Ollama library.

Create a Custom Model

# create_lucidai.py
import ollama

ollama.create(
    model='LucidAI',
    from_='llama3.2',
    system="Answer concisely and include only the most relevant information."
)

How to run:

python create_lucidai.py

Use the Custom Model

# ask_lucidai.py
import ollama

response = ollama.generate(
    model="LucidAI",
    prompt="Who discovered America?"
)
print(response['response'])

How to run:

python ask_lucidai.py

Delete a Model

# delete_lucidai.py
import ollama

ollama.delete("LucidAI")

How to run:

python delete_lucidai.py

In the next articles, you'll see how to build practical LLM-powered applications using these techniques!