Flora goes open, Gecko's Navy deal hits $300M
A data engine chewing through 200,000 hours of video a day on 500 GPUs just had its weights posted for free. Plus the Navy quadrupled a robot inspection contract and Eka came out with a force-first manipulation model.
Shown this week
Jim Fan's team open-sources Flora, its video-labeling data engine
Jim Fan released the weights for Flora, a suite of models that turns raw video into robot training signal, which he says matches Astra-level performance at 125x lower cost. It already runs in-house labeling 200,000 hours of video a day on 500 GPUs, producing timestamped captions, dense progress rewards and rubric scores.
It watches robot and human video and annotate what happened (timestamped instructions) and how well it went (rewards and failure reasons).
Eka raises $150M for a model that feels force, not just sees
Pulkit Agrawal announced a $150M Series A for Eka Robotics, built on a Vision-Force-Action model instead of the Vision-Language-Action models most labs use. He says force is the language of manipulation and that fast, reliable physical execution is the bottleneck, with the round led by a16z and F-Prime Capital.
Our robots retry until they succeed
Out in the world
Navy quadruples Gecko Robotics contract to $300 million
Gecko Robotics says its U.S. Navy contract has quadrupled to $300 million to improve fleet readiness. The company says its robots save up to $8 million in repair costs per ship and spot structural issues up to 50 times faster than traditional methods, cutting inspection times from months to days.
real robots doing real jobs for real customers, love that