Ollama

Ollama is a free, open-source tool for running open-weight LLMs locally with a single command, exposing a local API (including OpenAI-compatible mode) for applications. It emphasizes privacy, offline use, and no per-token cost, and offers an optional paid cloud for models that exceed local hardware. It has become a default building block for local AI development.

Ollama makes running open-weight LLMs on your own hardware simple. With one command you can pull and run models like Llama, Mistral, Gemma, Qwen, and DeepSeek locally, and it exposes a local API (including an OpenAI-compatible mode) so applications can talk to models without any external service. It runs on macOS, Windows, and Linux, and includes a desktop app alongside the CLI. The appeal is privacy, cost, and control: everything runs offline on your machine, so there are no per-token fees and no data leaving your device. Ollama has become a default building block for local AI, powering many developer tools, agents, and privacy-sensitive applications, and its model library and quantization handling make it easy to fit models to available memory. Ollama has added an optional cloud service for running larger models that exceed local hardware, with free, Pro, and higher tiers. The core local runtime remains free and open source. The main constraint is hardware: local model quality and speed are bounded by your CPU, GPU, and RAM, so the largest frontier-class models still favor the cloud.