How to Update Ollama and Its Models

Knowing how to update Ollama is one of those things every user eventually needs but nobody covers properly. There’s no built-in ollama update command — and the process is different depending on whether you want to update the Ollama application itself or the models you’ve downloaded. This guide covers both, on every platform.

Two types of update: the app vs your models

Before running any commands, it helps to know what you’re actually updating:

• The Ollama application — the binary or service that runs on your machine and serves the API. Updates here bring new features, bug fixes, and support for newer model formats.
• Your models — the AI models you’ve pulled (Llama, Gemma, DeepSeek, and others). These are updated separately using ollama pull.

Most confusion around updating Ollama comes from mixing these two things up. This guide covers both, step by step.

Step 1 — Check your current Ollama version

Before updating anything, check what version you’re running:

ollama --version

You can also query the version via the API:

curl http://localhost:11434/api/version

Compare the output against the latest release on GitHub Releases. If you’re already on the latest version, you may only need to update your models (see Step 3).

Step 2 — How to update Ollama

The update method depends on your operating system.

macOS

Ollama on macOS updates automatically in the background. When an update is available, you’ll see a notification in the menu bar. To install it:

1. Click the Ollama icon in the menu bar
2. Select Restart to Update

To manually trigger a check, quit Ollama from the menu bar and reopen it from Applications — it checks for updates on launch. If you installed via Homebrew, run:

brew upgrade ollama

Windows

Ollama on Windows also checks for updates automatically and will prompt you when one is available. To update manually:

1. Download the latest installer from ollama.com/download
2. Run the installer — it upgrades your existing installation in place
3. Your downloaded models are preserved automatically

If you used the step-by-step Windows installation guide to set up Ollama, this process is the same — the installer handles everything without you needing to uninstall first.

Linux

On Linux, Ollama doesn’t auto-update. Re-run the official install script to upgrade to the latest version:

curl -fsSL https://ollama.com/install.sh | sh

The script is safe to run on an existing installation — it replaces the binary and updates the systemd service definition without touching your models. After it completes, restart the service:

sudo systemctl restart ollama

Verify the service started cleanly:

sudo systemctl status ollama

To check the logs after updating:

journalctl -u ollama -n 50

If you followed the Linux installation guide when setting up Ollama, the update process is identical — the install script handles both the binary and the service file.

Docker

If you’re running Ollama in Docker, updating means pulling the latest image and recreating the container. Your models are stored in a named volume and will survive the update.

Pull the latest image:

docker pull ollama/ollama:latest

If you’re using Docker Compose:

docker compose pull
docker compose up -d

If you started the container with docker run directly:

docker stop ollama
docker rm ollama
docker run -d --gpus=all -v ollama:/root/.ollama -p 11434:11434 --name ollama ollama/ollama

The -v ollama:/root/.ollama flag is what preserves your downloaded models across container recreation — don’t skip it.

Step 3 — How to update your Ollama models

Ollama models are updated by pulling them again. There’s no automatic model update mechanism — Ollama will download any new layers since your last pull, so if only part of the model has changed it won’t re-download the whole thing.

Update a single model

ollama pull llama3.3

Replace llama3.3 with the name of the model you want to update. To see all models currently installed on your machine:

ollama list

This shows each model name, size, and when it was last modified.

Update all models at once

There’s no built-in ollama update --all command — this is the source of GitHub issue #2633, which has hundreds of comments from users looking for exactly this. The workaround is a one-liner:

Linux / macOS (bash):

ollama list | tail -n +2 | awk '{print $1}' | xargs -I {} ollama pull {}

Windows (PowerShell):

ollama list | Select-Object -Skip 1 | ForEach-Object { $model = ($_ -split '\s+')[0]; ollama pull $model }

Both scripts list all installed models, strip the header row, extract the model names, and run ollama pull on each one sequentially. If you have several large models, this will take time — run it when you’re not actively using Ollama for other tasks.

Installing a specific Ollama version

If you need to pin Ollama to a specific version — for example because a newer release broke something — use the OLLAMA_VERSION environment variable with the install script on Linux:

curl -fsSL https://ollama.com/install.sh | OLLAMA_VERSION=0.3.14 sh

Replace 0.3.14 with your target version. All available versions are listed on GitHub Releases.

One important note: the v0.4.0 release changed the internal model storage format. If you downgrade from v0.4.x to v0.3.x, models pulled under v0.4.x may not work and you’d need to re-pull them. Upgrading from v0.3.x to v0.4.x is a one-way migration — keep this in mind before going backwards.

Troubleshooting common update problems

Ollama service not starting after update (Linux)
Run journalctl -u ollama -n 100 to see the error. A common cause is old shared libraries left from a previous installation. Remove them and re-run the install script:

sudo rm -rf /usr/lib/ollama
curl -fsSL https://ollama.com/install.sh | sh
sudo systemctl restart ollama

Model behaving oddly after an app update
The template format that tells Ollama how to prompt a model can change between Ollama versions. Re-pulling the model usually fixes it:

ollama pull modelname

Not enough disk space to update models
Models are stored in ~/.ollama/models on Linux and macOS, and C:\Users\YourName\.ollama\models on Windows. Remove models you no longer use to free space:

ollama rm modelname

Pull fails with a network error
Ollama downloads are resumable. If a pull fails partway through, just re-run the same ollama pull command and it will continue from where it left off.

Should you always update Ollama?

Generally yes. Ollama releases are frequent and usually bring improved model support, performance gains, and bug fixes. Many newer model families — like Llama 4 and Gemma 4 — require a recent version of Ollama to run at all, so staying current means you can always pull the latest models.

The main reason to hold back is stability. If you’re running Ollama as a service that other applications depend on, test a new version before updating — and use the batch model update script during off-peak hours to avoid interrupting active requests.

If you’re new to Ollama, the complete beginners guide covers installation and first steps. Once you’re up to date, the best Ollama models guide is a good place to decide what to pull next.

Related Guides

• Ollama Thinking Mode: Enable, Disable and Control
• Ollama Security: Safe Remote Access with Tailscale