I wanted to create an art gallery featuring ITIA’s artwork in @odysseyml , but the model kept adding its own paintings. I was close to giving up when the artwork itself caught my attention. Apart from a few ITIA pieces in the reference image, all the other paintings were created by Odyssey’s AI, and some of them are genuinely amazing. Here’s the full video. Hope you enjoy it! If you want to try it check the first comment for prompt and photo. P.S. The neoclassical soundtrack, Tides of Glass, was also AI generated using my own prompt and the video as a reference.

Today we're launching Odyssey-3, the most powerful foundation world model yet. It sets a new state of the art on Physics-IQ, and powers robots, trains AIs, and generates interactive experiences. It's really cool. Experience the model today, all for free!

Yesterday Odyssey launched a new AI model called Odyssey 3. From what I understand, the idea is that todays AI is great with text and images, but still pretty bad at real world experience. Robots and machines need to practice how things actually move, bump into each other, break or react and doing all of that only in the real world is slow and expensive. This kind of model tries to simulate those experiences so robots can train with way less real data. They even showed examples of driving and controlling robots after relatively few hours of practice. I filmed a normal street in Madrid today because that is the messy reality these systems still have to deal with. Making it work outside, with people, traffic and surprises, is hard. Not sure yet how big this is (well yeah im sure), but it feels like one of the missing pieces if we want AI that can actually act in the physical world, not just talk about it. LLM + WM = SuperIntelligence

Odyssey-3 can generate interactive environments from a prompt, all in real time. It's a new kind of world simulator, and learns representations of physics, dynamics, and cause-and-effect from a broad dataset of visual observations.