Reflex-1
Decision models for agentsA 421M-parameter model for tool routing, intent classification and action selection. Public weights; runs on CPU or GPU.
Public weights
Ongoing investigations
Research
Decision models and robot learning. Results and methods from our current experiments.
We study models for local decisions and physical tasks. Our published work records its current results, hardware requirements and unfinished work.
Samsara-VLA
Robot learningRobot control with recurrent memory. 94.50% native LIBERO success in a 217M-parameter policy.
Experimental
VIO-lens-2
ResearchResearch in progress.
In progress