Introducing Agora-2, our next-generation multi-agent world model. Agora-2 supports up to 20 humans and agents interacting inside a shared environment, all simulated in real time. Our multiplayer research preview is available to try right now!

This research preview of Agora-2 enables up to 4 humans to join together to battle 16 agents, in an environment all simulated in real time by the world model. There's no game engine under the hood.

Going forward, we believe multi-agent world models will increasingly power important applications across AI training, AI safety, robotics, autonomous vehicles, defense, energy, cybersecurity, and gaming. Experience Agora-2 right now!

storing world state verifiably lets researchers audit how agents influenced one another

This is the next wave… I’m calling it now!

Can't wait to grief my little cousins' Minecraft world with 16 agents 🤣

Do it.

This is awesome. I want to check out that multiplayer research preview!

Let us know what you think.

Will do! :)

This looks like another level. A simulation model predicts how human and agent actions change the world, while a rendering model translates those changes into generated video for each player.

Now this is human computer interaction 🔥

Amazing. My main focus is now world models. Great work guys.

You are welcome ☺️

Kinda wild how far we’ve come from simple single-agent demos to 20 humans + AI agents sharing the same live world.

Too me back to my youth 😂

agents metaverse in reality this is awesome

That's right.

wow cool sounds like a room full of agents (love that Diablo game vibes in the video)

You guys are shipping at a crazy pace.

The team is doing amazing work.

@BacardiCapital it’s time to battle

I've been reading a lot of research on how agents coordinate. Agora-2 looks like a great testbed. It's a world model that generates a shared environment in real time where up to 20 humans and agents act together. Insanely cool!

👀

Love the Diablo references, you guys have good taste

no game engine under the hood is the kicker here. how are you handling the physics state consistency across all those agents in realtime

Congrats!!! :))

Thank you!

insanely cool

Thank you!

So glad that every one can enjoy this. This is the first step in the future of game dev.

Time to start shipping games lmao

cook

We are absolutely cooking.

The shared environment is the part that makes this really interesting.

Looks good, will definitely try it out!!!

Agents metaverse 👀

How long it takes until robots are in every home??

Yoo this is awesome

How will you measure collaboration quality as twenty agents share one world?

no game engine just a model hallucinating physics for 20 players simultaneously

great work!! (love your design btw)

Wait, I’m a total gaming newbie. If there’s no game engine underneath, how does it draw the scenes and characters in real time? Also, are these agents just fancy NPCs? What’s the actual difference?

20 agents in real-time is wild. Curious how you handle the latency bottlenecks on that scale.

I’ll definitely try this, Looks really interesting for multi-agent AI.

Finally, a game that tells you how many of the other players are bots.