Before a robot can learn a manipulation skill, someone must demonstrate it. OpenJaws records those examples without a robot: an operator moves a camera-equipped handheld gripper, and stereo vision reconstructs motion at metric scale. That avoids the absolute-scale ambiguity of monocular tracking.

The estimated parts cost is about $400. A benchmark contains 90 demonstrations across nine tasks, while offline stereo COLMAP reports 2.9 mm position error. Two UR10 policies reached 90% success each, but on just ten trials per task. Some promised code, hardware and dataset links still read Coming Soon, so the practical completeness of the open release remains to be checked.