jujishou

jujishou / codex-switch

Public

codex一键接入国产模型

16
2
69% credibility
Found May 28, 2026 at 16 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
Python
AI Summary

This is a Windows desktop tool that acts as a bridge between the OpenAI Codex App and other AI providers. Normally, Codex only works with OpenAI's official models, which can be expensive. This tool lets you swap in alternatives like DeepSeek, Kimi, Zhipu, Qwen, and others—often much cheaper—with one click. It works by running a local 'translator' service on your computer that converts Codex's requests into a format these other services understand. When you want to go back to the original setup, one click restores everything. The tool is self-contained (single executable), creates automatic backups before making changes, stores your account info separately for each provider, and stays local—only listening on your own computer.

How It Works

1
💡 You want to use Codex but find it expensive

You tried OpenAI Codex App but the official models cost too much, and you heard there are cheaper alternatives.

2
📦 You download a tiny tool that acts as a translator

The tool runs a local 'translator' on your computer that converts Codex's requests so other AI services can understand them.

3
🎯 You pick your favorite AI provider from a menu

The app shows you options like DeepSeek, Kimi, Zhipu, Qwen, and others—each with their own pricing and speed.

4
🔐 You enter your account information for that service

You paste your login details for the AI provider you chose, and the app saves them safely for next time.

5
You click the big button to start
🌟
First time switching

The app saves a backup of your original settings before making any changes

🔄
Coming back later

Your previous settings are remembered—no need to re-enter anything

6
✅ You open Codex and everything just works

Codex App now uses your chosen provider behind the scenes, and you see the same interface you always have.

🎉 You're saving money and talking to AI

Your conversations go through a cheaper provider, your results look the same, and you can switch back anytime with one click.

Sign up to see the full architecture

5 more

Sign Up Free

Star Growth

See how this repo grew from 16 to 16 stars Sign Up Free
Repurpose This Repo

Repurpose is a Pro feature

Generate ready-to-use prompts for X threads, LinkedIn posts, blog posts, YouTube scripts, and more -- with full repo context baked in.

Unlock Repurpose
AI-Generated Review

What is codex-switch?

codex-switch is a Windows desktop app that redirects OpenAI Codex App traffic away from OpenAI's official models to cheaper, faster Chinese alternatives like DeepSeek, Kimi, Zhipu, and Tongyi. It works by running a local proxy on your machine that translates Codex's API calls into standard OpenAI chat completions format, then forwards them to your chosen provider. The Python-based tool comes as a ready-to-run executable with a modern GUI -- pick a model, paste your API key, click start, and Codex talks to your preferred backend. One-click rollback restores your original Codex configuration when you want to go back.

Why is it gaining traction?

The hook is simple: Codex users can now access DeepSeek and Kimi at a fraction of OpenAI's pricing without leaving the Codex interface they already use. Built-in presets for nine popular platforms mean zero configuration for most users -- just select a template, enter your key, and go. The tool preserves your entire Codex config.toml, only modifying the model provider fields, so existing projects and plugins stay intact. Automatic backup before any changes and a single-button toggle between providers make experimentation risk-free.

Who should use this?

Cost-conscious developers running Codex who want to experiment with Chinese models without committing to OpenAI pricing. Teams evaluating different model providers for code completion tasks will find the quick-switch workflow useful. Windows users with existing accounts at DeepSeek, Kimi, or similar platforms are the primary audience -- the tool is explicitly Windows-focused, and non-Windows support is untested.

Verdict

With only 16 stars and a credibility score of 0.699999988079071%, this is a young, unproven project from an unknown author. The feature set is solid and the code shows reasonable structure, but limited community feedback means real-world edge cases may not be covered. Download the pre-built executable if you want to try it risk-free, but consider the lack of battle-testing before relying on it for anything critical.

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.