ResearchBenchmarksTeapot-Mug cross-category generalization on Real-WorldFollow14Success RateAffordGen-0.563.22710.78Apr 12, 2026Evaluation ResultsMethodMethodLinksSuccess RateAffordGenZero-shot=trueZero-shot=true2026.0414CPGenZero-shot=trueZero-shot=true2026.043DemoGenZero-shot=trueZero-shot=true2026.040