In this video I’m showcasing a way for you to use Claude Code completely for free by connecting it to Ollama and running open-source models locally. Claude Code is Anthropic’s agentic AI coding agent that lives in your terminal, and honestly I currently rank it as the best AI coding agent available right now. The key unlock here is that Ollama has Anthropic API compatibility, which means Claude Code can talk to open-source models like Qwen, GLM, Minimax, and more—hosted on your own machine—so you’re not paying for API usage. I walk you through the full setup: installing Ollama, installing Claude Code, picking a model your hardware can actually run, and then wiring everything up with the right environment variables. I also show the exact flow for launching Claude Code with an Ollama-backed model (so you don’t accidentally boot into the default Anthropic cloud models). The big takeaway: even if local model quality won’t always match official Claude, you still get Claude Code’s agentic workflow—scheduled tasks, sub-agents running in parallel, file discovery/code exploration, and research—powered by free local inference. I even demo building a clean landing page using a local Qwen 3.5 27B setup, which is honestly a gamechanger for the open-source community.