Ollama

Ollama is a small app that makes running local models almost too easy.

What it does:

1. You type ollama run llama3.1 (or qwen2.5, mistral, …).
2. It downloads a ready-to-use quantized build of that model (a few GB).
3. It runs it on your machine and exposes it on localhost — your laptop is now a private AI server.
4. Any tool that speaks the OpenAI-compatible API (chat UIs, code agents, scripts) can talk to it.

Why people use it:

• Private — your prompts and data never leave the machine.
• Free after download — no per-token billing, no rate limits.
• Offline — works on a plane, on a disconnected machine, anywhere.
• The standard dev sandbox for “what can I run on my hardware?”