Abstract representation of AI

Run ChatGPT-like AI models locally on your Mac or PC with Ollama.

published on 31 Mar 2025 in AI

As artificial intelligence (AI) continues to advance, large language models (LLMs) have become increasingly popular for their ability to understand and generate human-like text. However, running these models typically requires significant computational resources and internet connectivity. But what if you could run powerful AI language models locally on your Mac, without relying on cloud services or sacrificing performance? Enter Ollama, a straightforward tool that allows you to download and run LLMs like ChatGPT on your own computer.

Personal AI with local LLMs

Table of Contents

  1. Introduction

    • Explanation of the rise in popularity of large language models (LLMs)
    • Benefits of running AI models locally (computational resources, internet dependency)
    • Introduction of Ollama as a tool for running LLMs on personal computers

  2. What is Ollama?

    • Description of Ollama as an open-source application
    • Functionality of Ollama: downloading and running popular LLMs
    • Overview of models available for selection: Llama 3.2, DeepSeek R1, Phi-4, etc.

  3. Popular Models Overview

    • Brief descriptions of some key models available on Ollama:
      • Llama 3.2
        • Specs: 1 billion or 3 billion parameters
        • Ideal use cases: everyday tasks on standard computers
      • DeepSeek R1
        • Specs: 7 billion parameters
        • Strengths: solid performance with moderate resource demands
      • Phi-4
        • Specs: 14 billion parameters from Microsoft
        • Advantages: good balance of performance and resource usage
      • Llama 3.3
        • Specs: 70 billion parameters
        • Performance: strong, similar to larger models, requires powerful hardware

  4. Understanding Model Sizes

    • Importance of model size in selecting an LLM
    • Explanation of parameters as language understanding building blocks
    • Breakdown of model sizes:
      • Smaller Models (1B or 3B)
        • Run easily on typical computers
        • Disk space requirements: a few GBs
        • RAM requirements: around 16 GB
        • Best for everyday use
      • Larger Models (32B or 70B)
        • High computing power requirements
        • Increased disk space and RAM needs

  5. Setting Up Ollama on Your Mac

    • Step-by-step guide to installation:
      1. Download the executable
        • Source: official website
      2. Copy to Applications folder
        • Process of moving the executable
      3. Locate and start Ollama
        • Steps to find the Ollama icon
        • Instructions for one-time setup and installing CLI
      4. Open Terminal app
        • Command to run chosen model (e.g., run llama 3.2)

  6. Interacting with Your Local Language Model

    • How to use Terminal to engage with the local language model
    • Ability to ask questions and receive responses
    • Key advantage: offline interaction with the AI model

  7. Future Exploration

    • Mention of an upcoming video focusing on using local LLMs with a graphical user interface (GUI)

  8. Conclusion

    • Recap of the possibilities offered by Ollama for running AI models locally
    • Encouragement for viewers to explore and experiment with LLMs
    • Invitation to subscribe for more tutorials and engage in comments

What is Ollama?

Ollama is an open-source platform that provides a simple way to download and run popular LLMs on your local machine. With Ollama, you can choose from a variety of models, including Llama 3.2, DeepSeek R1, Phi-4, and more. These models are sorted by popularity and can be easily downloaded and installed on your Mac.

Overview of Popular Models

Before we dive into the installation process, let's take a brief look at some of the popular models available on Ollama:

  • Llama 3.2: A smaller model with 1 billion or 3 billion parameters, ideal for everyday tasks on standard computers.
  • DeepSeek R1: A reasoning model with 7 billion parameters, offering solid performance without huge resource demands.
  • Phi-4: A 14 billion parameter model from Microsoft that strikes a great balance between performance and resource usage.
  • Llama 3.3: The latest Llama model with 70 billion parameters, providing strong performance similar to larger models but requiring more powerful hardware.

Understanding Model Sizes

Number of parameters in large language models

When it comes to choosing an LLM, understanding the model size is crucial. Model sizes are measured in terms of parameters (e.g., 1B, 3B, 7B, or 70B), which can be thought of as building blocks for understanding language. More parameters generally mean better performance and more nuanced responses but also require more disk space and memory.

Here's a simple breakdown:

  • Smaller models (1B or 3B): Easier to run on typical computers, using less disk space (~ few GBs) and requiring around 16 GB of RAM. Ideal for everyday use.
  • Larger models (32B or 70B): Require significant computing power, typically needing more disk space and RAM.

Setting Up Ollama on Your Mac

Now that we've covered the basics, let's move on to setting up Ollama on your Mac:

  1. Download the executable: Download the Ollama executable from the official website.
  2. Copy to Applications folder: Copy the downloaded executable into your Applications folder.
  3. Locate and start Ollama: Locate the Ollama icon in the Applications folder, double-click it to start the one-time setup, and follow the prompts to install the command-line interface (CLI).
  4. Open Terminal app: Open the Terminal app and paste or type the command run llama 3.2 (or any other model you prefer). This will download and run the chosen LLM.

Interacting with Your Local Language Model

Once Ollama is set up, you can interact with your local language model using the Terminal app. Simply ask questions or provide input to the model, and it will respond instantly using its own training data – all without connecting to the internet!

In an upcoming video, we'll explore how to use local LLMs with a graphical user interface (GUI) similar to Anthropic's Claude app.

Conclusion

Running ChatGPT-like AI models locally on your Mac is now possible thanks to Ollama. With this beginner's guide, you've learned how to set up Ollama and start exploring the world of large language models. Whether you're a developer, researcher, or simply curious about AI, Ollama provides an exciting opportunity to experiment with cutting-edge technology right on your computer.

So go ahead, subscribe to our YouTube channel for more tutorials and updates, and join the conversation in the comments below!

Share this on

What do you think?