Ollama0.35.1

Run large language models directly on a Mac, using Ollama to generate text, assist with code, and automate work on the computer itself. It provides a local way to use open models when keeping prompts and other working data under your control matters. The software is aimed chiefly at developers who want local model support for coding agents, experiments, and everyday development tasks.

Ollama can connect open models to agents such as Claude Code and Codex, launching them with one command and allowing models to be switched without changing the surrounding workflow. Local models remain on the Mac, while prompts are never tracked or used for training; cloud models are hosted only in the United States, Europe, and Singapore. Open weights and open-source models are supported, and dedicated capacity is available for running several agents at once with consistent throughput.

Ollama runs on macOS 14 Sonoma or later and supports both Intel and Apple Silicon Macs. The current version is 0.35.1, and it is freeware. Local models are always free, while users can start with hosted usage and upgrade or add more usage when they need it; cloud access also offers the same open models with additional capacity.