Kyutai Releases REKI, a 2.4B Vision-Language-Action Model for Robot Manipulation

Kyutai, in collaboration with Inria and École des Ponts, has released REKI, a 2.4 billion-parameter vision-language-action (VLA) model designed for robot manipulation. Pretrained solely on public data, REKI demonstrates efficient memory usage and strong performance on out-of-distribution benchmarks, with all weights now available on Hugging Face.