#20 Time To Use Ollama
By now you’ve heard last week’s news about the AI startup DeepSeek releasing its open-source model DeepSeek-R1, which outperforms OpenAI’s ChatGPT-o1 using far fewer (and cheaper!) resources. The best part? It’s free to use.
Give it a try (https://chat.deepseek.com), and click on the DeepThink(R1) option to see its thought process!
If you want to know why DeepSeek is a game changer for AI, Mike Pound from the University of Nottingham summarises it perfectly in < 20 mins in this video.
Also last week, OpenAI announced a research preview of its AI agent called Operator, that is able to perform complex tasks in a web browser like booking trips, scheduling hotels, and more, with a high degree of autonomy.
We’re only one month into 2025, and there’s already a wave of AI innovations!
Regardless of which chatbot you use, the real excitement comes from having access to these open-source models (as opposed to OpenAI’s closed model) and leveraging them for our engineering and development work.
What if we could plug these models into our dev tools or run an offline ChatGPT?
If this sounds interesting to you, I’ll show you how to get started with Ollama.
What is Ollama?
Ollama is an open-source project that serves as a user-friendly interface for running Large Language Models (LLMs) on your computer.
Quick definition, LLMs are a Transformed-based architecture that is trained on large amounts of datasets and is good at predicting the next word based on the current text, and so on.
Ollama simplifies the process of downloading, installing, and interacting with a wide range of LLMs, making it relatively easy to use without relying on cloud-based platforms like AWS, Azure, GCP, etc.
How to use Ollama?
It’s very easy to get started.
Download the app https://ollama.com/download (ollama CLI will be available once installed)
Open your terminal, run
> ollama run llama3.2This will start a new chat with the llama3.2 model on your terminal where you can send in a prompt and get a response.
From there you can checkout DeepSeek’s R1 model.
Note: Make sure to download a small model with fewer parameters, say 1.5B, to avoid overloading your computer’s processor and causing it to freeze.
> ollama run deepseek-r1:1.5bHere’s a screenshot of using the CLI:
Ideally you won’t be using Ollama just to chat in the terminal. For a maximised developer experience, you have two options:
Use a web-based solution on top of Ollama.
Use an editor plugin on top of Ollama.
Let’s go through each one.
1. UI + Ollama
Having a web-based solution gives you the same ChatGPT-like experience—without internet connection—to generate results. There are several options for a web-based interface, but I recommend using Open WebUI.
https://github.com/open-webui/open-webui
Open WebUI provides a smooth setup and supports various LLM runners like Ollama. Check out their docs for more details.
To get started, install Open WebUI via Python and run the webui server locally:
> pip install open-webui
> open-webui serveOpen the browser (on http://localhost:8080) and from there you can select or download any models you want to use.
Here are other recommendations to check out:
2. Code Editor + Ollama
For a “GitHub Copilot”-like experience, you can use Continue.dev.
Continue is an AI tool that connects any running LLM to create custom autocomplete and chat experiences inside the IDE (available for VSCode and JetBrains).
Using Continue + Ollama will allow you to explain, fix, document, and modify any block of code in your editor. Just highlight the code and ask!
Conclusion
And that’s how you can use Ollama to run and integrate LLMs into your tools.
To wrap up, 2025 is going to be an incredible year for AI. I highly recommend following the news in AI and from companies like OpenAI and DeepSeek.
Let me know your thoughts if you’ve been using Ollama for a while.