GitHub radar
OpenCodex: An LLM For Claude Code & Codex.
A local proxy which can send Claude Code, Codex CLI, Claude Desktop, Grok Build requests to 40+ LLM backend(DeepSeek, Gemini, Ollama…)
OpenCodex is a local proxy that routes Claude Code, Codex CLI, Claude Desktop, Grok Build requests to any LLM backend providers(40+ supported by default, Any OpenAI compatible endpoint). It can be configured via web UI(localhost:10100). Supported backends include Anthropic, Google Gemini, xAI, Kimi, Azure OpenAI, Ollama, DeepSeek, Tencent Cloud… The provider/model format is supported in the model picker. Other functions such as: using a pool of ChatGPT /Codex accounts with session affinity/load balancing(keep active connections sticking to the same account), combos(a magic virtual model id which can fallback or round robin between multiple backends). OAuth supported backends include xAI, Anthropic, Kimi. Fully revert your settings via ocx stop.
Why a vibe-coder should care
Replace your low-quota Claude Code or try some cheaper/faster models without leaving your current app? One command away by OpenCodex! Works on all laptops(No GPU required).
How to install
Copy this and send it to your agent — Claude Code, Codex, any of them:
Install OpenCodex from https://github.com/lidge-jun/opencodex — run npm install -g @bitkyc08/opencodex && ocx start, open http://localhost:10100 and show me how to add a first LLM provider. Ask if you need API keys.
Runs on a regular laptop — no GPU needed; just Node.js 18+.
Open on GitHub▌ More finds