Reka AI · Omni Reasoning Model

Reka Rho-1

A 19B omni-reasoning model trained from scratch and released as a research preview on 5 October. One network understands and generates text, images and video, reasons over them and emits robot actions, all as tokens in a shared KV cache with no tool calls or second model. Each transformer block splits compute into an understanding stream and a diffusion generation stream. Reka reports base video generation at 0.79x real time, and a distilled Flash variant that cuts denoising from 99 steps to 8 and returns a 5.3-second clip in about one second, from its own testing. Video is capped at 672x384. No public weights or API; access to the preview is by request. Unrelated to Microsoft's 2024 Rho-1 language model.