Summary
Highlights
Claude Code is a powerful AI programming tool, but often comes with usage limits or requires payment. This video introduces OpenRouter, a platform that allows you to use Claude Code models for free and without limits, eliminating the need for powerful hardware or high RAM. The speaker, Álvaro Morales, CEO of Automatizado Hoy, emphasizes the importance of cost-efficiency in AI implementations.
Traditional free or local AI models are not scalable and often require significant RAM (e.g., 16GB), making them impractical for many users or expensive when hosted on a VPS. OpenRouter provides access to thousands of models, both free and paid, through a single API, offering scalability without demanding powerful hardware.
To get started, create an OpenRouter account and navigate to 'Explore Models'. Use the 'free' filter to find 28 available free models, many of which are powerful, like those from Nvidia Z.Aai and Queen models, which are highly recommended. These models have zero input and output token costs. Then, generate an API key from your profile settings.
Open your coding environment (e.g., VS Code), create a folder, and within it, a '.cloud' folder. Inside '.cloud', create a 'settings.json' file. Copy the project settings from OpenRouter's documentation into this file, remembering to define the model you'll use. Paste your previously generated OpenRouter API key into the 'settings.json' file.
OpenRouter requires adding a small credit amount (e.g., $5) to your account not for payment, but to authenticate you as a real user. Free models remain free. If a specific model (like a Queen model) returns an error due to saturation, you can configure 'settings.json' to use 'openrouter/free' as the model endpoint, allowing OpenRouter to automatically select any available free model, ensuring continuous and reliable use.
While free models are great for familiarizing yourself with Claude Code, remember that for complex projects, more powerful paid models like Opus 4.6 and Sonet 4.6 will yield better results. All resources mentioned in the video, including prompts and settings, are linked in the description. The video concludes with a prompt that allows Claude Code to automate the setup process.