Our verdict: Ollama is the standard engine for running open models on your own PC, and many other AI apps can use it as their backend. One command downloads and starts a model, and the local API works with OpenAI-style tools. Optional cloud models need an account, and big local models need a strong GPU.
About Ollama
Ollama is an open-source tool under the MIT license for downloading and running open language models. On Windows it installs as a native app in your user folder without administrator rights, runs in the background and adds the ollama command to the terminal. NVIDIA and AMD Radeon GPUs are supported for acceleration.
Models run locally and are served through an API on localhost port 11434, which many chat apps and coding tools can use. Ollama also offers cloud models that run on its own servers: they need sign-in, the free plan includes a limited selection, and most need a paid plan or credits. With cloud models your prompts leave your PC.
What it does well
- One-command models. Downloads and starts open models with a single ollama run command.
- Local API. Serves models on localhost for other apps, including OpenAI-compatible endpoints.
- GPU acceleration. Uses NVIDIA GPUs and AMD Radeon GPUs through ROCm or Vulkan.
- Desktop app. Includes a Windows app in the tray, not just a command-line tool.
- Optional cloud models. Runs larger models on Ollama servers after sign-in.
Who it is for
Good fit
Developers and tinkerers who want a reliable local model engine for chat apps, scripts and coding tools.
Look elsewhere
Users who want a full-featured chat interface with document libraries out of the box.
Pros and cons
Pros
- Free and open source
- Simple model download and run
- Local API used by many other apps
- No admin rights needed to install
Cons
- Basic interface compared with full chat apps
- Large models need a strong GPU and a lot of disk space
- Cloud models need an account and mostly a paid plan
How to install Ollama
Run OllamaSetup.exe; it installs to your home folder without admin rights and needs about 4 GB for the program. Update your NVIDIA or AMD driver first for GPU acceleration. Models can take tens to hundreds of gigabytes; set the OLLAMA_MODELS environment variable to store them on another drive.
Questions about Ollama
Is Ollama free?
Yes. Ollama is open source under the MIT license and running models locally is free. Its cloud models need sign-in; the free plan covers a limited set and most need a paid plan or credits.
Does Ollama work offline?
Yes for local models. After a model is downloaded, it runs on your PC and prompts stay on your computer.
