LOCAL AI RUNTIME

What is Ollama?

Ollama is a local AI runtime and model manager. It downloads model packages, loads them onto your computer and exposes a local API that other apps can use.

What does Ollama actually do?

It handles model downloads, local inference, model configuration and serving. Instead of manually managing model files and runtime arguments, you can pull a supported model by name and run it through one consistent interface.

Is Ollama the model?

No. Ollama is the software that runs models. Qwen, Gemma, Llama, gpt-oss and other families are the models themselves.

Does it need a GPU?

Not always. Ollama can use CPU processing and supported GPU acceleration. A GPU can make generation dramatically faster and determines which models can be fully loaded into VRAM.

Why do other apps connect to Ollama?

Ollama exposes a local API. Tools such as Open WebUI can use that API to give you a browser interface while Ollama continues doing the model work underneath.

Is Ollama private?

Running a model locally can keep inference on your machine, but web search, remote tools, cloud embeddings or third-party APIs can still send data elsewhere.

Official references: Ollama quickstart, GPU support and API documentation.