Don't waste 2 years learning to build LLMs like Claude & ChatGPT. deepseek's official setup switches your entire codex over to them, your existing models stop showing up at all. that's the actual problem codex-router solves. it adds models to the list instead of replacing anything. sol, grok, kimi, deepseek all sit in the same picker, you grab whichever fits the task. deepseek v4-flash at $0.28 per million output tokens versus opus 4.8 at $25, an 89x gap, without losing access to any model you were already using. http://github.com/duolahypercho/codex-router(CyrilXBT)