Paying per-message fees for every AI chat adds up fast, and sending internal documents to a cloud chatbot is not always comfortable. Ollama solves both problems by letting you download and run open AI models on your own computer. This tutorial walks you through installing Ollama, running your first model, and connecting it to your everyday tools.
What It Is
Ollama is a free, open-source tool for running large language models locally on macOS, Windows, and Linux. Instead of sending prompts to a hosted service, you download a model once and chat with it on your own machine. The project is published under the MIT License and, as checked on October 17, 2026, has 183,595 stars on GitHub (github.com/ollama/ollama). Its official description is: “Get up and running with Kimi, GLM, MiniMax, DeepSeek, gpt-oss, Qwen, Gemma and other models.” The model library at ollama.com/library lists the models you can pull, including Gemma, Qwen, DeepSeek, Llama-family models, and others.
Who It’s For / SMB Use Case
Ollama fits small businesses that want AI help without a monthly per-seat bill or without sending sensitive text to a third party. Practical uses include:
- Private drafting: Draft emails, product descriptions, and replies on your own machine, so client details never leave your office.
- Offline work: Keep working when the internet is slow or down, because the model runs locally after download.
- Low-cost experimentation: Test AI workflows before committing to a paid platform. The software itself is free; your only cost is the computer you already own.
- Developer building block: Developers can point apps, chat interfaces, and coding assistants at Ollama through its REST API or its Python and JavaScript libraries.
Ollama is not a replacement for a polished hosted chatbot for everyone. It has no built-in fancy interface beyond the terminal, and answer quality depends on which model you pick and how powerful your computer is. But for cost control and privacy, it is one of the simplest starting points in open-source AI.
Installation
Ollama supports macOS, Windows, and Linux, plus Docker. Pick the method below that matches your system. All commands come from the official Ollama README.
macOS
Install with the terminal command:
curl -fsSL https://ollama.com/install.sh | sh
Or download the installer manually from ollama.com/download/Ollama.dmg.
Windows
Install with PowerShell:
irm https://ollama.com/install.ps1 | iex
Or download the installer manually from ollama.com/download/OllamaSetup.exe.
Linux
Install with the terminal command:
curl -fsSL https://ollama.com/install.sh | sh
For a manual Linux install, follow the instructions at docs.ollama.com/linux.
Docker
The official Docker image ollama/ollama is available on Docker Hub. If you already run Docker, check the official docs for the run command that matches your GPU setup, because GPU flags differ between NVIDIA and CPU-only hosts.
How to Use Ollama
Step 1: Start Ollama and run your first model
Open a terminal and run:
ollama run gemma4
This downloads the Gemma 4 model on first use and drops you into a chat prompt. Type a question, press Enter, and the model answers in the terminal. To see the full model list, visit ollama.com/library. For more detail, see the official quickstart guide.
Step 2: Launch a coding or assistant integration
Ollama can connect to coding assistants and chat apps. To launch a specific integration, use the ollama launch command. For example, to launch Claude Code:
ollama launch claude
To turn Ollama into a personal AI assistant across messaging apps, the README gives this example:
ollama launch openclaw
Supported integrations listed in the README include Claude Code, Codex, Copilot CLI, DeepSeek Harness, Droid, and OpenCode. Check the CLI reference for the full list and options.
Step 3: Call Ollama from your own app with the REST API
Ollama exposes a local REST API. This example from the README sends a chat request:
curl http://localhost:11434/api/chat -d '{
"model": "gemma4",
"messages": [{
"role": "user",
"content": "Why is the sky blue?"
}],
"stream": false
}'
See the API documentation for all endpoints. A small business developer can use this to add AI features to an internal tool, such as summarizing support tickets or drafting replies, without paying per-call fees to a hosted provider.
Step 4: Use Ollama from Python or JavaScript
Install the Python library:
pip install ollama
Then chat with a model in Python, using the example from the README:
from ollama import chat
response = chat(model='gemma4', messages=[
{
'role': 'user',
'content': 'Why is the sky blue?',
},
])
print(response.message.content)
For JavaScript, install the package with npm i ollama and import it as shown in the README. If you prefer a graphical chat window instead of the terminal, the README lists community chat interfaces such as Open WebUI, LibreChat, and AnythingLLM. Check each project’s own docs before installing.
Pricing / Limitations
Ollama itself is free and open source under the MIT License. There is no per-message charge from Ollama, because the model runs on your hardware. That does not mean the total cost is zero: you need a reasonably capable computer, disk space for model downloads (models can be several gigabytes), and electricity. Larger, more capable models need more RAM and preferably a dedicated GPU, so a basic office laptop may only run smaller models comfortably. Answer quality also varies by model; small local models will not match the most powerful hosted models on hard reasoning tasks. Finally, Ollama is a model runner, not a finished business app: you or your developer must wire it into the tools your team actually uses. If the README does not state something clearly, check the official docs at docs.ollama.com.
Bottom Line for SMB
Ollama is the cheapest, most private way for a small business to start using AI. Install it in minutes, run a model locally, and test real tasks such as drafting, summarizing, and answering questions from your own documents. If you outgrow it, you can still keep it as the local engine behind a nicer interface or a custom tool. For broader context on how AI agents fit into a small business, read our guide to AI agents explained for small business and our roundup of the best AI writing tools for small business.
Frequently Asked Questions
Does Ollama need an internet connection?
You need internet to install Ollama and download a model. After the model is downloaded, you can chat with it locally without an internet connection.
Is Ollama free for business use?
Yes. Ollama is published under the MIT License. You still pay for your own hardware, and any third-party tools or hosted models you connect may have their own costs.
Which model should a small business start with?
Start with a small, well-known model from the official library, such as Gemma 4 as shown in the README example, and test it on your real tasks. If answers are too weak, try a larger model or check the official docs for guidance.
