Reka AI's omni-model Rho-1 handles text, images, video, and robot control in a single model

via The Decoder · 1 views · 5h ago

One AI model now controls text, video, and robots. Reka AI just unveiled Rho-1, a 19-billion-parameter 'omni-model' that's making waves. This isn't just another language model; it's designed to be truly universal. Rho-1 uniquely processes and generates multiple modalities. We're talking text, images, video, and even robot control actions — all within a single neural network. This unified approach could simplify complex AI deployments significantly. What’s even more impressive is its efficiency. Reka AI trained Rho-1 on just 320 H100 GPUs in about three months. That's a fraction of the compute typically required for models with comparable breadth, hinting at new possibilities for scaling. This breakthrough pushes us closer to truly versatile AI agents that can perceive and act across diverse digital and physical environments with unprecedented coherence and efficiency.

← Back to the feed