OpenAI and ModRetro Released a Codex-Built Game Console for $200, and the First Batch Sold Out in Under a Day
OpenAI and the startup ModRetro have released Chromatic: Codex Edition, a Game Boy Color clone for games written by the Codex AI agent. You describe the idea in plain language, Codex works out the logic, draws the sprites and animates the characters, and a plugin compiles everything into a finished ROM. You can test the game in a browser and on the console, then write it to a real cartridge. For developers, it is a clear example of vibe coding that ends with something you can hold in your hands.
AI-processed from TG: @hiaimedia (Hi, AI • Новости технологий); edited by Hamidun News
Vibe coding usually ends with a file in a repository or a page in a browser. OpenAI decided to take this chain one step further, to a physical object: together with the startup ModRetro, the company presented ModRetro Chromatic: Codex Edition. This is a version of Chromatic, a modern Game Boy Color clone, built for games written by the Codex AI agent. The console costs $200, and the first batch sold out in less than a day.
How an idea becomes a cartridge
The process is short. A user describes the concept to Codex in plain language, for example a platformer about a cat that collects fish. The agent builds the game logic, draws the sprites, and animates the characters. A special plugin then compiles everything into a finished ROM. A ROM is a single file containing the game that the console can run.
From there, the author has three options. They can open and test the game in a browser, run it on the console itself, or write it to a real cartridge. That last step is what sets the project apart from typical AI coding demos: the result ends up in your hands, and you can give it to a friend or put it on a shelf.
The console is only partly tied to AI-made games. It also works with regular cartridges for the Game Boy and Game Boy Color, so it can serve as a modern replacement for the old handheld.
Where the hype came from
The device was first shown at OpenAI's DevDay presentation in late September. Several thousand consoles were handed out to conference attendees then. Social media was so impressed by the gadget that resellers began listing it on eBay for about $2,000. After that level of interest, OpenAI decided to release the console at scale.
The new version costs $200. The first batch sold out in less than a day, and the next sale opens on November 19. The authors of the Telegram channel Hi, AI joke that on that day gamers will probably switch over to GTA VI. That is a joke, and there is no sales data behind it.
What developers and product managers should take away
First: tight constraints help the agent. The Game Boy Color has a small screen, simple graphics, and a clear ROM format. Under these conditions, Codex can more easily carry a task through to a result than it can in a large project with dozens of services. For learning and experimentation, this is a convenient sandbox: you can test an idea in an evening.
Second: this is a product move. Instead of a slide with benchmarks, OpenAI demonstrates what the model can do through an object people want to buy and show to others. Collector interest, eBay prices, and the rapid sellout of the batch all work as marketing for Codex.
Third: the scheme can be repeated in other niches. A closed platform, a clear build format, and a plugin that turns the agent's work into a finished artifact form a workable template. The same path could suit widgets, editor extensions, or small mobile games: the agent writes the code, and the build tool takes it the rest of the way to a running product.
Limitations
The source does not make clear how complex the games Codex produces are, or how many fixes a person needs after the first run. Most likely these are small games in the spirit of classic handheld consoles, and full-scale projects are unlikely here. The result depends on how precise the description is and how carefully the author checks the game in the browser before writing it to a cartridge.
Also, $200 is a considerable sum for a device that mainly serves as a showcase for the agent. It is worth buying out of curiosity and for a collection, but it is unlikely to become a working tool. Still, the signal for the industry is clear: AI agents are moving from text prompts to finished digital and physical products. Developers would do well to look for narrow platforms where this path can be followed end to end, from idea to finished artifact.
- Nov 21, 2026California Sues OpenAI: Its AI Agents Escaped Control at Least 80 Times
- Nov 21, 2026OpenAI GPT-6 Cyber Found About 35 Flaws in iPhone, Mac and Other Apple Devices
- Nov 21, 202622 Claude Agents Spent Six Weeks on Black Hole Problems — and Solved Two
- Nov 21, 2026Google to split Gemini chat history from model training, so you can opt out without losing your history
Key Questions Answered
How an idea becomes a cartridge?
The process is short. A user describes the concept to Codex in plain language, for example a platformer about a cat that collects fish. The agent builds the game logic, draws the sprites, and animates the characters. A special plugin then compiles everything into a finished ROM. A ROM is a single file containing the game that the console can run. From there, the author has three options. They can open and test the game in a browser, run it on the console itself, or write it to a real cartridge.
Where the hype came from?
The device was first shown at OpenAI's DevDay presentation in late September. Several thousand consoles were handed out to conference attendees then. Social media was so impressed by the gadget that resellers began listing it on eBay for about $2,000. After that level of interest, OpenAI decided to release the console at scale. The new version costs $200. The first batch sold out in less than a day, and the next sale opens on November 19.
What developers and product managers should take away?
First: tight constraints help the agent. The Game Boy Color has a small screen, simple graphics, and a clear ROM format. Under these conditions, Codex can more easily carry a task through to a result than it can in a large project with dozens of services. For learning and experimentation, this is a convenient sandbox: you can test an idea in an evening. Second: this is a product move.
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.
The AI world, distilled — once a week
Seven stories that actually mattered, hand-picked. No noise, no reposts, no press releases.
Done! Check your inbox for a confirmation.