TREE NEWS update: Skild AI has open-sourced Steel, an Apache-licensed physics solver for robot training that runs inside the Newton simulator. The company reports the solver reaches up to 13 times the simulation throughput of PhysX in mixed-task tests. The release gives robotics developers a faster open tool for training robot policies in simulation.
Skild AI Open-Sources Steel Physics Solver for Robot Training
The strategic weight here sits less in the raw speed claim than in the licensing: an Apache-licensed solver inside Newton lowers the cost of entry for teams that would otherwise build or license simulation infrastructure. The benchmark is a company-reported, mixed-task comparison against PhysX, so the 13x figure describes one test configuration rather than a general verdict. Robotics developers training policies in simulation are the immediate beneficiaries, though simulator incumbents now face a free alternative in the same stack. Whether independent groups reproduce the throughput claim, and whether Steel attracts outside contributors, is the open question.
Generated by AI for reference only.
Share on WeChat
Open WeChat → Scan → then tap "…" to send to a chat or Moments.
Tap "…" in the top-right corner to send to a chat or share to Moments.