Ollama is an open-source tool for running open models on your own machine. You install it on macOS, Windows or Linux, or use the official Docker image, and pull models from its library, such as gpt-oss, Qwen, Gemma and DeepSeek. Ollama provides a local REST API and client libraries for Python and JavaScript, so applications and developer tools can work with it directly. There is also an optional cloud service in which models run on Ollama's servers.
Ollama is made by Ollama Inc. in San Francisco in the United States and is released under the MIT licence. If you run models locally, prompts and responses stay on your own infrastructure. The cloud models are different: Ollama says it hosts them primarily in the United States and may route requests to Europe and Singapore during high demand. According to Ollama, prompt and response data is not logged or used for training. A European organisation that wants to keep data under its own control should therefore use Ollama locally.