I was tired of switching tabs to compare ChatGPT, Gemini & Copilot. So I built a <1MB Chrome extension to run them all in parallel.

starwing — Sat, 04 Apr 2026 13:07:58 +0000

hey guys! 👋

I wanted to share a side project I recently open-sourced: EasyChat.

As a developer who relies heavily on AI, I found myself constantly frustrated by a very specific problem: when tackling a tricky issue, I often want to ask ChatGPT, Gemini, and DeepSeek the exact same question to compare their answers. But hopping between 3 different browser tabs, copy-pasting the prompt, and losing track of historical context felt incredibly clunky. Oh, and burning through expensive API quotas wasn't ideal either.

So, I built a lightweight Chrome Side Panel extension to solve this once and for all. It's under 1MB.

GitHub: https://github.com/starwingChen/easyChat (It's fully open-source, I'd really appreciate a star if you find it useful! ✨)

🚀 Why I think you'll like it

EasyChat lets you chat with multiple top-tier AI models side-by-side right from a Chrome sidebar.

1. 🤖 100% Free Web-scraping + API Support (No paywalls or hidden limits)

This is the best part: ChatGPT, Gemini, Perplexity, and Copilot are simulated through web requests. So as long as you are logged into their web interfaces in your browser, the extension piggybacks off your active session. This means zero API cost, and it doesn't drain your API credits.
It also supports standard API key configs for models like DeepSeek and Qwen if you prefer the API route.

2. ⚡ True Parallel Answers & Multi-Layout

You type one prompt, and the extension broadcasts it to all active bots simultaneously. It supports draggable layouts (1 grid, 2 vertical, 2 horizontal, 3 grid, 4 grid). Put ChatGPT on the left, Gemini on the right, and drop DeepSeek at the bottom. You can instantly spot which one is hallucinating.

3. ⌨️ Snappy & Distraction-Free

Keyboard Shortcut: Alt+J toggles the sidebar instantly. No need to click extension icons.
History Snapshots: Your parallel sessions are saved as read-only snapshots, so you can always look back at your comparisons.
Zero Bloat: It's insanely fast and does exactly what it needs to do without the fluff.

🛠️ For the nerds (Here's how I made it)

If you're a dev interested in MV3 or AI coding, you might find the repository interesting. The codebase was 100% written by an AI agent! I personally only focused on the overarching architecture design and reverse-engineering the authentication APIs (my design docs are in .rules).

Tech Stack: React 19 + Tailwind CSS v4 + Vite, built for Chrome Manifest V3. Kept it lightweight with SCSS and Radix-UI headless components rather than heavy UI libraries.
Architecture design: Strictly uses Domain-Driven Design concepts. All bots (web-based or API) are abstracted under a generic BaseBotAdapter so the UI layer never cares about network protocols.
Web Simulation bypasses: For complex auths like Copilot, it uses MV3's declarativeNetRequest to dynamically rewrite WebSocket headers.

I built this simply because I wanted to "ask a quick question" without breaking my workflow. If you think this workflow could help you, give it a try!