ollama
Download a model, run it on your own machine, and talk to it offline.
What it is
Ollama is a single program that downloads, runs and serves large language models on your own computer. You install it once, pull a model with one command, and from then on the model lives on your disk — no API key, no per-token bill, no prompt leaving your machine.
Under the hood it wraps llama.cpp and handles the parts that usually trip people up: model formats, GPU offloading, memory sizing and a local HTTP server. That server speaks the same shape of API as the hosted providers, so code you already wrote against a cloud model can often be pointed at localhost instead.
It is the piece most people install first when they want to try a private AI setup, because it is the shortest path between "I have a computer" and "a language model is answering me".
Why we recommend it
The install is genuinely one command
Most self-hosted AI stacks are a weekend project. Ollama is one installer and one pull command. That difference is the whole reason it spread — it removes the setup step that kills most self-hosting attempts.
Your prompts never leave the machine
Useful for client work, medical or legal text, internal documents, or anything you would rather not paste into someone else's cloud. This is a privacy property you cannot get from an API, at any price.
Hardware support is broad
Ships builds for Windows, macOS (including Apple Silicon), and Linux on both amd64 and arm64, with ROCm and Jetson variants published alongside. You are unlikely to own a machine it has no answer for.
It stays out of your way
There is no dashboard to learn. Models are managed by name from the command line, and the background service just keeps a local endpoint warm for whatever app you point at it.
How it scores
These are our editorial ratings, not benchmark results. We score on how the project behaves in practice — how long it takes to get running, how good the documentation is, how responsive the maintainers are, and how well it holds up over time. The licence and star count above are read from GitHub and are a snapshot.
Pros and cons
Strengths
- One-command install on all three desktop platforms
- Works fully offline once a model is pulled
- OpenAI-compatible local API, so existing code ports easily
- Very active development and a large model library
- No account, no telemetry requirement, no cost per request
Trade-offs
- Large models need a lot of RAM or VRAM — check your hardware before pulling a 70B model
- The Windows installer is around 1.5 GB, and models are much larger still
- Fewer tuning knobs than running llama.cpp directly, which matters if you need fine control
How to get it
Direct download
Clicking a button downloads the file straight from the project's release — you stay on this page.
One-line install
Prefer the terminal? Copy a command and you are done.
curl -fsSL https://ollama.com/install.sh | sh
winget install Ollama.Ollama
Source and documentation
At a glance
| Category | Local AI |
|---|---|
| Licence | MIT |
| Written in | Go |
| GitHub stars | 180,549 |
| Repository | ollama/ollama |
Licence and star count were read from the GitHub API when this page was generated and are a snapshot. Check the repository for the current figure and the authoritative licence text.
Other Local AI tools
We are not affiliated with ollama or its maintainers. We list it because it is genuinely useful and genuinely free. If you plan to build a commercial product on top of it, read the licence on the repository first.