Ollama

What is Ollama?

Ollama gives developers a way to run open AI models on their own computers and to use hosted models when local hardware is insufficient. Its downloads cover macOS, Windows and Linux, while cloud models can also be called from application code through an API. Local execution is useful for experimenting with models without paying a hosted inference charge, but speed and usable model size depend on the computer’s GPU and memory. The cloud alternative avoids downloading model weights and uses metered credits, with concurrency and included usage varying by plan. Developers choosing between the two should compare hardware capacity with expected token use rather than treating a paid subscription as unlimited cloud inference.

Run open models locally or use cloud inference, with desktop downloads and a code-facing API. Local hardware limits and cloud token charges require separate planning.

Capabilities and use cases

Official sources

We link to the original product source so you can verify current availability, pricing, and features.