miuuyy/codex-chatgpt-web

Use ChatGPT Web (including Pro) as a native model in Codex — with context, tools, streaming and images, without using Codex quota.

What it solves

This project provides a bridge that allows users to use the ChatGPT Web interface (including Pro accounts) as native models within the Codex environment. It eliminates the need for paid API keys by leveraging the web-based version of ChatGPT, while maintaining the native Codex user interface, context lifecycle, and tool harness.

How it works

The software operates as a local bridge (Responses bridge) that routes Codex tasks to a fresh ChatGPT Temporary Chat session. It uses an embedded browser to automate the interaction with the ChatGPT web page, streaming reasoning, tool activity, and Markdown responses back into Codex.

Depending on the mode selected, it can function in three ways:

  • Browser-only: Basic routing of prompts and responses.
  • Full harness: Uses the OpenAI tunnel-client and MCP (Model Context Protocol) to connect ChatGPT's tool calls back to the Codex task's filesystem, shell, and other configured tools.
  • Zero Risk: A manual mode where the user pastes prompts and responses to avoid automation risks.

Who it’s for

Developers and Codex users who want to access high-tier ChatGPT Web models (like Pro) without paying for API fees, or who want to integrate the native web-based reasoning and tool capabilities of ChatGPT into their existing Codex workflow.

Highlights

  • Native Integration: ChatGPT Web models appear directly in the Codex model picker.
  • Full Tool Access: Via MCP, the web-based model can interact with the local filesystem and shell of the active Codex task.
  • Session Management: Uses Temporary Chats to maintain task-bound sessions and supports native Codex compaction at context boundaries.
  • Cross-Platform Launcher: A dedicated app for macOS, Windows, and Linux that handles sign-in, setup, and health checks.
  • Fail-Closed Design: Explicitly errors out if models or tools are missing rather than silently failing or switching capabilities.

Related

  • Project
    totec448-spec/chat-on-steroids

    Chat On Steroids is an Electron desktop app + Chrome extension that lets ChatGPT call local tools (file read/write, patches, shell commands, screenshots, mouse/keyboard control) and spawn additional worker chats. It stores every tool call locally, supports session hand‑off (Compact & Resume), and enforces per‑folder permissions. Installers are provided for Windows/macOS/Linux; a Secure MCP tunnel and a restricted OpenAI API key are required.

  • Project
    XiaoDuoYa/codex-with-chatgpt

    A secure bridge that lets the ChatGPT web app act as the planning and review brain for Codex coding sessions, reducing API token consumption.

  • Project
    composio-community/open-chatgpt-atlas

    Open ChatGPT Atlas is an open‑source Chrome/Edge extension (and optional Electron app) that lets an AI control the browser (via Gemini 2.5 Computer‑Use) and call over 500 third‑party services (via Composio). Install with Node, load the built `dist` folder, add your Google and optional Composio API keys, then use the sidebar chat to give natural‑language commands like “click this button” or “create a GitHub issue.” It offers visual feedback, safety confirmations, and runs entirely client‑side.

  • Dispatch
    Codex in ChatGPT: AI Coding Agents for Linux and Desktop

    OpenAI has integrated Codex into the ChatGPT desktop app and released a Linux version, providing a multi-agent coding environment with IDE and CLI extensions.

  • Project
    kardolus/chatgpt-cli

    ChatGPT CLI is a cross‑platform command‑line client for many LLM providers. It offers streaming queries, thread‑based chat history, prompt‑file injection, multimedia I/O, and an experimental agent mode (ReAct / Plan‑Execute) that can run shell/file/web‑search tools under budget and policy constraints. It also supports the Model Context Protocol (MCP) for custom tool calls. Install via Homebrew or a pre‑built binary, configure API keys in `~/.chatgpt-cli/config.yaml`, and start chatting with `chatgpt "your question"` or `chatgpt --interactive`.