Ollama

A runtime between model weights and applications

Ollama packages model downloads and inference behind a command-line interface, an API and a desktop chat application. Its terms identify the provider as Ollama Inc. This is software for using models from different developers, rather than a claim that Ollama trained every model in its catalog. [1][4]

The July 2025 desktop launch brought native chat applications to macOS and Windows. It exposed model selection, file inputs and context controls without requiring terminal interaction. Image understanding and large context windows still depend on the chosen model and the machine's available resources. [2]

Ollama's later account includes both local inference and cloud models. Readers choosing a deployment therefore need to distinguish where a particular request executes. The runtime's MIT license also remains separate from the licenses and usage conditions attached to individual model weights. [3][4][5]