Chrome extension ChatGPT Multi Pane mimics Codex logic without the limits or the extra cost

A developer has published ChatGPT Multi Pane, a Chrome extension that runs several ChatGPT windows in parallel, each with its own role. One writes code, another reviews it, a third holds the context, and when a task is done, a notification arrives in Telegram. Under the hood it is the same gpt-5.5 that powers Codex, but without Codex's limits and pricing. The repository is open source on GitHub.

AI-processed from Habr AI; edited by Hamidun News
Chrome extension ChatGPT Multi Pane mimics Codex logic without the limits or the extra cost
Source: Habr AI. Collage: Hamidun News.
Listen to article
Article Fact Sheet & Key Figures
ParameterDetails
Primary sourceHabr AI ↗
Editor-in-Chief
Key metric #1the first conversation writes code for the task
Key metric #2the second checks the result and looks for errors
Key metric #3the third holds the overall context of the project or task

A developer from Habr has published ChatGPT Multi Pane, an experimental Chrome extension that adds several parallel conversation panes to the ChatGPT.com interface. In how it works, it reproduces the key features of OpenAI's Codex, but without a paid subscription and without hard limits.

The problem the extension solves

Codex is OpenAI's flagship agentic tool for development. The model can write code on its own, run it, check the results and iterate. But Codex has two serious drawbacks: rising cost and hard usage limits. According to the extension's author, with every update the limits get smaller and the price goes up.

Under the hood, Codex runs on gpt-5.5 with a particular system prompt. That means regular ChatGPT is the same model, and with the right organization of tasks you can get a similar result. The developer realized this and decided not to wait for OpenAI to ease its terms.

How Multi Pane works

The extension adds several independent panes to the ChatGPT.com interface, each working as a separate conversation with gpt-5.5. Each pane is assigned its own role:

  • the first conversation writes code for the task
  • the second checks the result and looks for errors
  • the third holds the overall context of the project or task
  • the fourth accumulates a queue of next steps

While all the conversations run in parallel, the user can do other things instead of sitting and waiting for the model to finish a response. When a task is done, the extension sends a notification to Telegram. This approach is especially useful when refactoring a large project or handling several independent tasks at once.

"One conversation on the left wrote code, another on the right checked

the result, a third held the context, and meanwhile I could sketch out a queue of next steps and go off for a cup of tea" — the extension's author, on Habr.

Technical details

The extension is built as an open-source Chrome plugin; the repository is available on GitHub under the name chatgpt-multi. It needs no third-party APIs, extra keys or special subscriptions: it uses the same session as regular ChatGPT in the browser.

Installation is standard: download the repository and load it as an unpacked extension through Chrome's developer mode. The Telegram notification integration is configured through the Telegram Bot API, which any developer can handle.

The limitations are the same as those of ChatGPT itself: several parallel requests may slow each other down on a non-premium plan. The author calls the project experimental and warns of possible problems when the interface is updated, since the extension works on top of the page's DOM elements.

The multi-agent approach as a trend

Running several agents with different roles in parallel is one of the key architectural ideas in modern LLM development. Large systems are built this way: one agent plans, another executes, a third verifies. OpenAI is developing Codex in this direction, Anthropic is building similar capabilities into Claude, and Google into Gemini.

The key advantage of this architecture is specialization without mixing contexts. When a single LLM holds the entire task in one conversation, it loses the thread over long responses. Splitting by role lets each conversation stay short and precise. ChatGPT Multi Pane is a homemade implementation of the same idea on top of the public interface: without the memory and tools of paid platforms, but the principle works.

What this means

It is another example of the community reproducing expensive paid features with accessible tools. Developers are increasingly taking OpenAI's products apart and assembling analogues from whatever is at hand. If Codex doesn't fit your budget or its limits don't suit you, a multi-pane ChatGPT with Telegram notifications is a viable alternative for simple tasks.

Story

Key Questions Answered

What are the key details on The problem the extension solves?

Codex is OpenAI's flagship agentic tool for development. The model can write code on its own, run it, check the results and iterate. But Codex has two serious drawbacks: rising cost and hard usage limits. According to the extension's author, with every update the limits get smaller and the price goes up. Under the hood, Codex runs on gpt-5.5 with a particular system prompt. That means regular ChatGPT is the same model, and with the right organization of tasks you can get a similar result. The developer realized this and decided not to wait for OpenAI to ease its terms.

How Multi Pane works?

The extension adds several independent panes to the ChatGPT.com interface, each working as a separate conversation with gpt-5.5. Each pane is assigned its own role: the first conversation writes code for the task the second checks the result and looks for errors the third holds the overall context of the project or task the fourth accumulates a queue of next steps While all the conversations run in parallel, the user can do other things instead of sitting and waiting for the model to finish a response. When a task is done, the extension sends a notification to Telegram. Thi

What are the key details on Technical details?

The extension is built as an open-source Chrome plugin; the repository is available on GitHub under the name chatgpt-multi. It needs no third-party APIs, extra keys or special subscriptions: it uses the same session as regular ChatGPT in the browser. Installation is standard: download the repository and load it as an unpacked extension through Chrome's developer mode. The Telegram notification integration is configured through the Telegram Bot API, which any developer can handle. The limitations are the same as those of ChatGPT itself: several parallel requests may slow eac

ZK
Zhemal Khamidun — CPO at Alpina Digital, product manager of AlpinaGPT. Building AI products since 2019. Hamidun News is a pet project that grew into a production aggregator with 6,000+ readers.
✦ Hands-On AI Engineering · Hamidun Academy

Build & Ship Production AI Agents Without Boilerplate Code

Start your free 72-hour trial at Hamidun Academy (academy.hamidun.com) — interactive lessons on Claude Code & Cursor, live Tech Radar, and production agent templates.

Start 72h Free Trial →Vibe-Coding Intensive
What do you think?
Loading comments…