Reka AI's omni-model Rho-1 handles text, images, video, and robot control in a single model
Reka AI previewed Rho-1, a 19B omni-model for text, images, video, and robot control.
Reka AI released a research preview of Rho-1, a 19-billion-parameter model that processes and generates text, images, video, and robot-control actions in one network. Modalities are tokens in a shared context window, without tool calls or external models, and video generation can continue as new instructions arrive. The same weights predict camera images and drive robot movement; an inverse-dynamics model extracted control signals from ordinary internet videos. Training used 320 H100 GPUs over about three months. Reka previously shipped multimodal Reka Core in April 2024.
- Rho-1 is a 19B model handling text, images, video, and robot actions.
- All modalities share one context window, with no external model calls.
- It continues video generation while accepting new instructions.
- Training used 320 H100 GPUs for about three months.
- Organizations
- Reka AI
From the original article
1 min read at the-decoder.com
Reka AI's Rho-1 is a 19-billion-parameter omni-model that processes and generates text, images, video, and robot control actions in a single neural network. Trained on 320 H100 GPUs in about three months, it uses a fraction of the compute today's top models need. Instead of routing tasks to specialized systems, Rho-1 runs all modalities as tokens in one shared context window.
By Matthias Bastian · the-decoder.com
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.