Explore Ollama Models: Download, Run, Switch Guide

Posted on by

You installed Ollama. You ran your first model, maybe Llama 2. Now you want to see what else is out there and explore Ollama Models.

Ollama supports many different AI models. These models have unique abilities and sizes.

This guide shows you how to explore Ollama models. You will learn to find, download, list, run different, and remove models.

Mastering these simple commands unlocks Ollama’s full power on your computer.

Where to Explore Ollama Models: The Model Library

First, you need to know where to look for models. The official Ollama model library is the best place.

Visit the library at ollama.com/library. This website lists many available models.

You will see names like mistral, codellama, or llama2. These are model names.

Models also have “tags.” A tag often shows the model size or version. Examples are mistral:7b or llama2:13b.

Look at the model descriptions. They tell you about the model’s capabilities. Some are good for coding, others for writing.

Downloading Models with ollama pull

You can download models using a simple command. This command is ollama pull.

Running ollama run [model_name] automatically downloads the model if you don’t have it. But ollama pull just downloads it without starting a chat.

Use ollama pull to download models before you need them. This saves time later.

To download a model, you need its name or tag. Let’s download the mistral model as an example.

Open your terminal or command prompt. Type the command below and press Enter.

Command: Download Mistral Model

ollama pull mistral

The command starts the download. You will see progress as it downloads parts of the model. This might take a while depending on the model size and your internet speed.

Ollama verifies the download when it finishes. It tells you when the download is complete.

Listing Your Downloaded Models with ollama list

As you download more models, you might forget which ones you have. Ollama has a command to show you.

The command is ollama list. It lists all models stored on your computer.

Open your terminal again. Type the command and press Enter.

Command: List Installed Models

ollama list

The output shows a list. Each line is a model you have downloaded. You see the model name, its tag, how big it is, and when you last used it.

Here is what the output might look like:

    NAME          ID            SIZE    MODIFIED
llama2:latest 78e26419b446  3.8 GB  2 hours ago
mistral:latest c5f221591137  4.1 GB  About a minute ago

This output confirms which models are ready to use locally.

Running Different Models with ollama run (Revisited)

You already know the ollama run command. You used it for your first chat.

This command also lets you start a chat with any *downloaded* model. You just add the model’s name or tag after the command.

If you downloaded mistral using ollama pull, you can now run it quickly.

Open your terminal. Type the command for the model you want to use and press Enter.

Command: Run Mistral Model

ollama run mistral

If the model is already downloaded, Ollama loads it fast. You will see the prompt ready for your input, just like with Llama 2.

You are now chatting with the Mistral model. To exit the chat, type /bye and press Enter.

Switching Models

Switching between models is very easy in Ollama. You do not need a special “switch” command.

To use a different model, simply exit your current chat session (type /bye). Then, run the ollama run command again, but use the name or tag of the model you want to switch to.

Each ollama run [model_name] command starts a new session with the specified model.

Removing Models with ollama rm

AI models take up a lot of disk space. You might download models to try them out.

If you don’t use a model anymore, you can remove it. This frees up space on your computer.

The command to remove a model is ollama rm. Use it carefully.

Important: Removing a model is permanent. You will need to download it again if you want to use it later.

Find the name of the model you want to remove using ollama list first. Let’s say you want to remove llama2.

Open your terminal. Type the command below and press Enter.

Command: Remove Llama2 Model

ollama rm llama2

Ollama will remove the model files. It usually doesn’t ask for confirmation. The command runs and finishes quickly if successful.

Run ollama list again after removing. You will see that the removed model is no longer in your list.

You can also remove multiple models at once. Just list their names after the command, like:

ollama rm model1 model2 model3.

Choosing the Right Model for You

Many factors affect which model is best. Model size is a big one. Larger models (like 70B) often perform better but need more system resources (GPU, RAM).

Smaller models (like 7B or 13B) run faster on less powerful hardware. They are great for getting started or for tasks that don’t need high complexity.

Consider the model’s purpose. Some models are trained for specific tasks, like coding (e.g., codellama).

Experimentation is key. Try different models from the Ollama library to see what works best for your needs and computer.

Troubleshooting Common Issues

Sometimes things don’t work perfectly. Here are solutions for common Ollama model issues.

If you get an “Error pulling” a model, check the model name spelling. Also, make sure you have a stable internet connection and enough disk space for the download.

If you get an “Error running” a model, check if you downloaded it successfully. Use the ollama list command to verify it’s installed.

If you have trouble removing a model with ollama rm, double-check the model name or tag using ollama list. Ensure you are typing it exactly right.

FAQs

How much space do models need?

Model size varies greatly. Smaller ones like 7B can be around 4GB. Larger ones like 70B can be over 40GB. Always check the size on the Ollama library page before downloading.

Can I download models manually?

Ollama is designed to manage downloads with the ollama pull command. This is the standard and easiest way to get models into Ollama’s system.

How do I update a model?

To get the latest version of a model, just run ollama pull [model_name] again. Ollama will check for updates and download the newest version if available.

Conclusion

You now have the essential commands to manage models in Ollama. You can find new models, download them, see what you have, run different ones, and remove them when needed.

The ollama pull, ollama list, ollama run, and ollama rm commands are your tools.

Explore the Ollama library and try out different AI models. Discover their unique capabilities. Find the models that best suit your projects and interests.

Managing your local LLMs is straightforward with these commands.