explaingit

nicosuave/opencodex

Analysis updated 2026-08-16 · repo last pushed 2026-08-02

Audience · developerComplexity · 3/5ActiveSetup · moderate

TLDR

opencodex is a local proxy that lets you use any AI model, Claude, Gemini, DeepSeek, Grok, or local Ollama models, inside coding tools like Codex CLI, Claude Code, and Grok Build that normally lock you into one provider.

Mindmap

mindmap
  root((repo))
    What it does
      Proxies AI model requests
      Works with many coding tools
      Supports 40+ providers
    Key features
      Local dashboard for setup
      Account pooling and routing
      Subagent delegation
      History-safe injection
    Tech stack
      Node 18+
      macOS Linux Windows
      No containers needed
    Use cases
      Cost control across models
      Mix models per task
      Local model integration
      Multi-account management
    Audience
      Developers wanting flexibility
      Teams with cost concerns

Code map

Detail Auto

An interactive map of this repo's files and how they connect — its source is parsed live in your browser. Click Visualize to build it.

filefunction / class

What do people build with it?

USE CASE 1

Use a cheaper local model for routine code generation and switch to a premium reasoning model only for complex architecture decisions.

USE CASE 2

Route multiple ChatGPT or Codex accounts through one proxy so sessions automatically go to the least-used healthy account.

USE CASE 3

Use DeepSeek or Grok inside Claude Code without waiting for official provider support.

USE CASE 4

Run a locally hosted model through Ollama inside OpenAI's Codex CLI for fully private coding assistance.

What is it built with?

Node.jsTypeScriptCodex CLIClaude CodeClaude DesktopGrok BuildOllama

How does it compare?

nicosuave/opencodex000madz000/rfid-attendance00kaku/gallery-slider-block
Language—TypeScriptJavaScript
Last pushed2026-08-022024-07-222021-05-19
MaintenanceActiveDormantDormant
Setup difficultymoderateeasyeasy
Complexity3/52/52/5
Audiencedeveloperdevelopergeneral

Figures from each repo's GitHub metadata at analysis time.

How do you get it running?

Difficulty · moderate Time to first run · 30min

Requires Node 18+ and API keys or OAuth login for each AI provider you want to use.

The README does not include license information for this project.

In plain English

opencodex lets you use any AI model you want, Claude, Gemini, DeepSeek, Grok, or even a locally hosted model through Ollama, inside developer tools that normally lock you into one provider. It works with OpenAI's Codex CLI and app, Claude Code, Claude Desktop, and Grok Build. Instead of waiting for those tools to officially support your preferred model, you just point them at this proxy and pick whatever model fits your task. At its core, opencodex sits between your coding tool and the AI provider, translating requests so they speak the right language to whichever model you've chosen. You install it, run a quick setup command, and a local dashboard opens where you add your providers and API keys (or log in via OAuth for some). Once configured, your chosen models show up directly in the model pickers of Codex, Claude Code, and the other supported tools as if they were native options. You select a model using a simple "provider/model-name" syntax, and everything else, streaming, tool calls, reasoning, images, works normally in both directions. This is useful for developers or teams who want flexibility and cost control. For example, you might use an expensive reasoning model for complex architecture decisions but switch to a cheaper, faster model for routine code generation. You can even set up subagent delegation, where one model handles a task and hands a subtask to a different model. The project also includes account pooling for ChatGPT/Codex accounts, if you have multiple accounts, it tracks their usage quotas and automatically routes new sessions to the least-used healthy account, while keeping ongoing conversations pinned to the account that started them so nothing breaks mid-task. The project runs natively on macOS, Linux, and Windows with no complex container setup, requiring only Node 18 or higher. It supports 40+ built-in providers out of the box and is notably designed to be "history-safe", it injects itself into Codex without remapping your existing chat history, so an unclean shutdown won't scramble your past conversations. The README doesn't go into deep detail on every edge case, but the breadth of supported tools and providers is extensive.

Copy-paste prompts

Prompt 1
Install opencodex and configure it to route Claude Code through DeepSeek's API instead of Anthropic, using the provider/model-name syntax.
Prompt 2
Set up opencodex account pooling with two ChatGPT accounts so new coding sessions automatically route to the account with the most remaining quota.
Prompt 3
Configure opencodex to expose a local Ollama model inside the Codex CLI model picker so I can code with a fully offline AI model.
Prompt 4
Create a subagent delegation setup in opencodex where a reasoning model handles architecture questions and delegates routine code generation to a faster, cheaper model.

Frequently asked questions

What is opencodex?

opencodex is a local proxy that lets you use any AI model, Claude, Gemini, DeepSeek, Grok, or local Ollama models, inside coding tools like Codex CLI, Claude Code, and Grok Build that normally lock you into one provider.

Is opencodex actively maintained?

Active — commit in last 30 days (last push 2026-08-02).

What license does opencodex use?

The README does not include license information for this project.

How hard is opencodex to set up?

Setup difficulty is rated moderate, with roughly 30min to a first successful run.

Who is opencodex for?

Mainly developer.

Open on GitHub → Explain another repo

This repo across BitVibe Labs

Verify against the repo before relying on details.

Alternatives & similar repos