Free Claude Code

Simply put, it’s an "AI request proxy". When you normally use Claude Code, your requests go straight to Anthropic’s servers. This tool runs a local proxy that quietly redirects those requests to whichever AI model you prefer—like free local models via Ollama, DeepSeek, or NVIDIA APIs—while keeping the Claude Code client completely unaware and fully functional.

It tackles the pain points of expensive official API costs, usage limits, or the desire to run local models for free. Instead of juggling multiple clients or writing complex forwarding scripts yourself, this proxy standardizes different model APIs into a format Claude Code understands. You get to switch models freely within the same tool, saving money and gaining flexibility.

• •Developers and students who want to use Claude Code but find official subscriptions too pricey or quota-restricted
• •Tinkerers who already run local LLM environments (like Ollama) and want to leverage their own hardware
• •Programmers who want to seamlessly switch between multiple AI models for coding assistance within one terminal or IDE

• •Running local open-source models via Ollama to experience Claude Code’s generation and debugging features at zero cost
• •Routing complex reasoning tasks to powerful models like DeepSeek/Opus while sending simple completions to cheaper ones, optimizing costs
• •Sending voice notes or text in Discord, letting the proxy automatically call models to write code or debug in the background

• •Plug-and-play: Just set environment variables and run. It perfectly supports Claude Code’s official /model switcher and tool-calling protocols
• •Smart routing: Route traffic by model name without touching code, letting different task complexities hit different AI services
• •Multi-platform support: Beyond terminal and VS Code, it includes Discord/Telegram bot wrappers and voice transcription, offering open-ended workflows

Requires basic command-line familiarity. You’ll need Python 3.14 and the uv package manager installed. You must configure a .env file with an API key for your chosen provider (or run locally with Ollama, which costs nothing but requires your own GPU/CPU). No coding required—just run a few commands to start the proxy.