Ollama
Description
A program for running open AI models on your own computer, with an option to use models in the cloud. It works on macOS, Windows and Linux, and there is an official Docker image. You download a model and chat with it from the command line (for example with the ollama run command), or connect your own apps through a local REST API; client libraries exist for Python and JavaScript. Ollama can run models on the CPU alone, so a graphics card is not required; with a supported GPU the model loads into video memory. The project is open source under the MIT license and builds on llama.cpp. It has no model of its own: you pick one from its library (Qwen, Gemma, DeepSeek, gpt-oss and others) and many other tools can plug into it. Prompts are not stored or used for training, according to the site (checked 25 October 2026).
Pros and cons
Pros
- Free and open source (MIT) for running on your own computer
- macOS, Windows, Linux and Docker
- Works without a graphics card, on the CPU
- Local API and libraries for your own apps
Cons
- The main way to work is the command line and API; a chat window is secondary
- Large models need a lot of memory, and the site gives no minimum requirements
Key facts
- Free access
- Yes Running models on your own computer is free; cloud models have usage limits and paid plans
- Pricing
- Pricing on the official site →
- Developer
- Ollama
- Checked on
- Oct 25, 2026