ResearchBenchmarksLightbulb Screw on Real-world 1.0 (test)Lightbulb Screw on Real-world 1.0 (test)Follow100Success RateSCORE (DSRL)-4235077Jun 25, 2026Evaluation resultsMethodMethodLinksSuccess RateSCORE (DSRL)variant=DSRLvariant=DSRL2026.06100SCORE2026.06100Base2026.0650Res-RL2026.0650FPO2026.0637.5RialTo2026.060