How to Run Ollama Models Locally and Access Them Over a Network - Part 2

In Everything a Developer Should Know About Ollama - Part 1, we built the Ollama mental model. Now let’s use it like a developer: install it, run models, call the API, move model storage, and expose it safely on your network.

⏱️ Time to Complete

Around 10-15 minutes.

🎯 What you’ll achieve / learn

• Install Ollama and confirm the CLI works
• Run models from terminal and UI
• Call Ollama from your own app using the local HTTP API
• Store large model files on a custom disk
• Expose Ollama to your LAN safely
• Know the common endpoints developers check first

⚙️ Install Ollama

Download Ollama from the official download page:

https://ollama.com/download

On Windows and macOS, install the app from the download page. On Linux, the official install command is:

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

After installation, check that the CLI works:

ollama --version

Start the local server manually if needed:

ollama serve

Most desktop installs start the Ollama background service automatically.

💻 Run models from the terminal

Find models in the Ollama library, then run one by name.

ollama run gemma4

Pull without immediately chatting:

ollama pull gemma4

List downloaded models:

ollama ls

Remove a model:

ollama rm gemma4

See which models are currently loaded:

ollama ps

Stop a running model:

ollama stop gemma4

Run a one-shot prompt:

ollama run gemma4 "Explain dependency injection in one paragraph."

Use a multimodal model with an image, if that model supports vision:

ollama run gemma4 "What is in this image? C:\Users\me\Desktop\screenshot.png"

Generate embeddings:

ollama run embeddinggemma "Hello world"

🖼️ Run models from a UI

There are a few UI paths.

First, Ollama's desktop app may be enough for normal local chatting if your platform/install includes the app UI. Install Ollama, open the app, choose or enter a model, and start chatting.

Use the CLI for advanced commands like pull, rm, create, ps, and custom Modelfile workflows.

Second, you can use a web UI on top of Ollama. Popular choices include:

• Open WebUI
• AnythingLLM
• LibreChat

The usual flow is:

1. ✅ Install and start Ollama
2. ⬇️ Pull a model with ollama pull <model>
3. 🖥️ Start the UI
4. 🔌 Configure the UI to use Ollama at http://localhost:11434
5. 🤖 Select the model inside the UI

For example, if a UI asks for the Ollama host/base URL, try:

http://localhost:11434

If it asks for the API base URL:

http://localhost:11434/api

🔌 Call Ollama from your own code

Ollama exposes an HTTP API, so you can test it with curl before writing code.

The official local API base URL is:

http://localhost:11434/api

Generate endpoint:

curl http://localhost:11434/api/generate -d '{
  "model": "gemma4",
  "prompt": "Why is the sky blue?"
}'

Chat endpoint:

curl http://localhost:11434/api/chat -d '{
  "model": "gemma4",
  "messages": [
    { "role": "user", "content": "Explain REST APIs to a junior developer." }
  ]
}'

List local models through the API:

curl http://localhost:11434/api/tags

Check version:

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

The API also includes endpoints for embeddings, model details, pulling, creating, copying, pushing, deleting, and listing running models. See the official Ollama API docs.

If you are building an app, you can connect this API to stacks like Next.js, Node.js, Python FastAPI, LangChain, or LlamaIndex.

💾 Store models in a custom location

Ollama models can become large very quickly. If your system drive is small, move them to a bigger disk.

Default model locations from the official FAQ:

OS	Default model path
macOS	~/.ollama/models
Linux	/usr/share/ollama/.ollama/models
Windows	C:\Users\%username%\.ollama\models

To use another folder, set the OLLAMA_MODELS environment variable.

Windows PowerShell for the current shell:

$env:OLLAMA_MODELS = "D:\ollama-models"
ollama serve

Windows permanent user environment variable:

[Environment]::SetEnvironmentVariable("OLLAMA_MODELS", "D:\ollama-models", "User")

Then quit and restart Ollama from the Start menu.

Linux with systemd:

sudo systemctl edit ollama.service

Add:

[Service]
Environment="OLLAMA_MODELS=/mnt/ai/ollama-models"

Then reload:

sudo systemctl daemon-reload
sudo systemctl restart ollama

On Linux, make sure the ollama service user can read and write to that directory:

sudo chown -R ollama:ollama /mnt/ai/ollama-models

🌐 Expose Ollama to your network

By default, Ollama binds to 127.0.0.1:11434. That means only your own machine can reach it.

To expose it to other devices on your LAN, set OLLAMA_HOST:

OLLAMA_HOST=0.0.0.0:11434 ollama serve

On Windows PowerShell for the current shell:

$env:OLLAMA_HOST = "0.0.0.0:11434"
ollama serve

On Linux systemd:

[Service]
Environment="OLLAMA_HOST=0.0.0.0:11434"

Then:

sudo systemctl daemon-reload
sudo systemctl restart ollama

From another machine on the same network, test:

curl http://YOUR_HOST_IP:11434/api/version

Or:

curl http://YOUR_HOST_IP:11434/api/tags

Common endpoints to remember:

Purpose	Endpoint
Health/version check	GET /api/version
List local models	GET /api/tags
Generate text	POST /api/generate
Chat messages	POST /api/chat
List loaded models	GET /api/ps

Security warning: do not expose Ollama directly to the public internet without protection. Put it behind a reverse proxy, firewall, VPN, tunnel with access controls, or an auth layer. Tools like Nginx, Cloudflare Tunnel, Tailscale, and WireGuard are common options depending on your setup.

If a browser app or extension cannot call Ollama because of CORS/origin rules, configure OLLAMA_ORIGINS. For example:

OLLAMA_ORIGINS=http://localhost:3000,chrome-extension://* ollama serve

Only allow the origins you actually need.

⚡ Quick command cheat sheet

# Run a model
ollama run gemma4

# Download a model
ollama pull gemma4

# List downloaded models
ollama ls

# List loaded/running models
ollama ps

# Remove a model
ollama rm gemma4

# Stop a model
ollama stop gemma4

# Start the server manually
ollama serve

# Create a custom model from a Modelfile
ollama create dev-helper -f ./Modelfile

# API checks
curl http://localhost:11434/api/version
curl http://localhost:11434/api/tags

🔗 Useful links

• Ollama download
• Ollama model library
• Ollama CLI reference
• Ollama API documentation
• Ollama FAQ: model paths, environment variables, network exposure
• Docker
• Open WebUI
• Cloudflare Tunnel