Advanced AI Technologies
Reka AI previews multimodal model with robot-control capabilities
Reka AI released a research preview of Rho-1, a 19-billion-parameter multimodal model combining text, images, video, and robot control.
Reka AI released a research preview of Rho-1, a 19-billion-parameter model designed to handle text, images, video, and robot control in one system. The model is positioned as a multimodal system with applications extending beyond text generation.
Its ability to process multiple data types could support applications that combine language, visual understanding, video analysis, and interaction with physical machines. Robot control introduces an embodied-AI dimension that connects model outputs to real-world action.
The release is described as a research preview, meaning the technology is presented for early evaluation rather than as a mature production product. Its capabilities therefore represent a direction for development rather than evidence of broad commercial deployment.
The announcement reflects the industry’s effort to consolidate multiple modalities and robotics capabilities into unified models, potentially reducing the need for separate systems for language, vision, video, and control.
Original source: AI Daily Post