Claude Code 100% GRÁTIS com Ollama (Sem Pagar NADA)

You can run Claude Code for free, without an API key, without a Pro plan, without paying anything. Using an AI model running directly on your machine. In this video, I install everything live and create a project from scratch to prove it works. 🔴 Vibe Coding Training (Antigravity, Claude Code and more): https://app.horadecodar.com.br/lp/for... 🟪 Hosting for n8n that I recommend: https://hostinger.com.br/matheusbattisti (use the coupon HORADECODAR to get an extra 10% discount) 📘 Prompt Engineering Guide: https://app.horadecodar.com.br/ebookp... Claude Code is Anthropic's coding tool that runs in the terminal. Normally, you need a $20 per month Pro plan or API credits to use it. But Ollama changed that. Ollama is a runtime that runs AI models locally on your machine. Since version 0.14, it has implemented compatibility with the Anthropic API. This means that Claude Code thinks it's talking to Anthropic's servers, but in reality, it's talking to a local model running on your computer. The model used is the Zhipu AI GLM-4.7 Flash. It's a Mixture of Experts model with 30 billion parameters, but only activates 3 billion per token. In other words: it runs fast on consumer hardware. It has 128k context and supports tool calling, which is essential for Claude Code to function properly. To run the GLM-4.7 Flash locally, you need reasonable hardware. The recommended setup is an NVIDIA GPU with at least 12GB of VRAM, such as an RTX 3060 12GB, 4070 Ti, 4080, or 4090, or a Mac M-series with 32GB of unified RAM. The minimum is 16GB of total RAM. You can run it on the CPU without a GPU, but it will be much slower. Installation is in 4 steps. First, install Ollama from the ollama.com website. Second, download the template using `ollam pull glm-4.7-flash`. The template is around 18GB. Third, install Claude Code if you don't already have it. Fourth, run `ollam launch claude --model glm-4.7-flash` and you're done. Ollama automatically configures the environment variables and launches Claude Code connected to the local template. No API key, no paid plan, nothing. In the demo, I create a complete landing page for a fictional startup called AutoBot. HTML, CSS, and JavaScript are separated. Dark mode with gradient, animated counters, cards with flip-on-hover functionality, pricing with monthly and annual toggle that recalculates via JS, FAQ with accordion, smooth scroll. All generated by Claude Code running 100% locally. Then I iterate, adding a contact form with JS validation to show that the iteration works. I'm honest about the limitations. The local model is not the same as Sonnet or Opus. It works well for static websites, scripts, isolated functions, simple CRUD. It struggles in complex projects with multiple interconnected components. The speed: a task that Sonnet solves in 30 seconds can take 3 to 5 minutes in the local model. For beginners, for smaller projects, for offline work, or for keeping code private, it's excellent. For professional work on large projects, the Pro plan is still worthwhile. If your machine can't handle local execution, Ollama offers cloud models that run on their servers for free. Run `ollama pull glm-4.7:cloud` and then `ollama launch claude --model glm-4.7:cloud`. You need internet, but it's free. 🔗 Video Links: ▸ Ollama: https://ollama.com Join our Discord server and follow me on social media: 🟣 Discord Hora de Codar:   / discord   🔴 Instagram:   / horadecodar   🔷 Telegram: https://t.me/horadecodar TIMESTAMPS 00:00 Free Claude Code with Ollama 00:45 Step-by-step guide to making Claude Code free with Ollama 03:45 Downloading Ollama 04:10 Choosing an LLM model to run Claude Code for free 07:00 Using Claude Code for free with a local LLM 08:50 Using Claude Code for free with a cloud-based LLM 10:55 My thoughts on Claude Code + Ollama = free