SimplerEnv Google Robot (Variant Aggregation) leaderboard

The same three Google Robot tasks as visual matching, averaged over scene variants: backgrounds, lighting, table textures, distractors and camera poses. It tests robustness rather than real-to-sim fidelity, and is reported separately from visual matching. Scores are the three-task mean, recomputed from per-task values when a paper averages in a fourth task.

Simulation · Google Robot (RT-1 / Fractal) · SAPIEN (ManiSkill2)Report an error
84.9%Best · CORAL
22Models ranked
Near saturationSaturation
Nov 18, 2026Numbers checked
#ModelTypeScoreDateSource
1CORALFrontier RoboticsVLA84.9%, self-reported by the authors, not a leaderboard entryMar 2026
2RLDX-1RLWRLDVLA82.3%, self-reported by the authors, not a leaderboard entryMay 2026
3Xiaomi-Robotics-0Xiaomi RoboticsVLA77.4%, self-reported by the authors, not a leaderboard entryFeb 2026
4SimVLAFrontier RoboticsVLA76.2%, self-reported by the authors, not a leaderboard entryFeb 2026
5EO-1Shanghai AI LabVLA76.1%, self-reported by the authors, not a leaderboard entryAug 2025
6X-VLATsinghua AIR et al.VLA75.7%, self-reported by the authors, not a leaderboard entryOct 2025
7HumeSJTU & Shanghai AI LabVLA74.1%, self-reported by the authors, not a leaderboard entryMay 2025
8StarVLAStarVLA TeamVLA73.8%, self-reported by the authors, not a leaderboard entryApr 2026
9MolmoActAi2VLA72.1%, self-reported by the authors, not a leaderboard entryAug 2025
10MemoryVLATsinghua UniversityVLA70.8%, self-reported by the authors, not a leaderboard entryAug 2025
11SpatialVLAShanghai AI LabVLA67.5%, self-reported by the authors, not a leaderboard entryJan 2025
12CogACTMicrosoft Research AsiaVLA66.2%, self-reported by the authors, not a leaderboard entryNov 2024
13RT-2-XGoogle DeepMindVLA65.6%, reported by a third partyMay 2024
14DitaShanghai AI LabPolicy65.3%, self-reported by the authors, not a leaderboard entryMar 2025
15π0-FASTPhysical IntelligenceVLA59.0%, reported by a third partyAug 2025
16π0.5Physical IntelligenceVLA58.1%, reported by a third partyJun 2026
17RT-1Google (Robotics at Google, Everyday Robots)VLA57.4%, reported by a third partyMay 2024
18π0Physical IntelligenceVLA54.8%, reported by a third partyAug 2025
19TraceVLAUniversity of MarylandVLA49.1%, self-reported by the authors, not a leaderboard entryDec 2024
20OpenVLAStanford / UC Berkeley et al.VLA40.0%, reported by a third partyJan 2025
21RT-1-XGoogle DeepMindVLA36.9%, reported by a third partyMay 2024
22OctoUC Berkeley (RAIL)VLA1.6%, reported by a third partyMay 2024

Metric: Variant-aggregation success, mean of 3 Google Robot tasks (pick coke can, move near, open/close drawer). One result per model: RoboFYI re-run, then official leaderboard, then paper, then third party. Runs that deviate from the official protocol are listed on the model page but not ranked.

Wrong or missing number? Report an error