Artificial Intelligence News

First AI-generated racing game with multiplayer mode goes open-source

New Delhi: Gamers and developers scrolling through tech Twitter this week might’ve paused on something wild: a fully AI-generated multiplayer racing game. Yep, it’s real. Built by Enigma Labs, the game is called Multiverse, and it’s not just a visual demo, it’s playable, real-time, supports multiple players, and yes, it’s open-source.

This might be the first time we’re seeing a proper AI-generated multiplayer world, where players can race together inside an environment shaped and rendered in real-time using a diffusion AI model. The best part? You don’t need a supercomputer to run it. According to the devs, “Training and research cost < $1.5K. Run it on your own PC.”

Built by AI, shaped by players

Multiverse lets players interact and influence the game world together. Unlike traditional racing games where the environment is fixed and pre-built, here the track, visuals, and even elements of movement are dynamically created by AI using diffusion models, a kind of generative system that usually powers text-to-image platforms like Stable Diffusion.

The experience isn’t just about racing, it’s about being part of a living, changing, AI-simulated world. Think GTA meets Midjourney, but you’re not just watching it happen, you’re in it, steering through it in real-time with friends.

The multiplayer layer, which was always the missing piece in AI gaming demos, finally clicks here.

Everything is open for devs

Enigma Labs posted, “We are open-sourcing everything: code, data, weights, architecture, and research.” That means you can poke around the project, fork it, mod it, or even build something entirely different on top of it. It’s all out there, free to explore.

This approach could shake things up for indie game developers, modding communities, and even students trying to break into game design without million-rupee GPUs or expensive 3D tools.

The dev cost is less than ₹1.27 lakh

You read that right. The entire system, including training, cost less than $1,500 (about ₹1.27 lakh). That’s less than what some folks spend on a custom gaming PC alone. And it runs on standard gaming rigs, no cloud servers or massive GPU farms needed.

It’s rare to see something this experimental and playable at the same time. And it feels like a clear sign that generative AI isn’t just for weird face filters or fake voices anymore. It’s quietly entering the game dev scene in ways that are getting harder to ignore.

What’s next?

Hard to say. Could be more modders jumping on board. Could be full-on AI racing tournaments in a few months. For now, Multiverse feels like one of those rare drops where AI, code, and gameplay actually line up into something fresh.

Hugginface: https://huggingface.co/Enigma-AI

Github: https://github.com/EnigmaLabsAI/multiverse