SimplerEnv Google Robot (Visual Matching) leaderboard
Real-to-sim replica of the RT-1 Google Robot setup, with the simulated scene overlaid on real backgrounds to match the real camera view. Policies are usually fine-tuned on the RT-1 (Fractal) data. Scores are the mean of three tasks; many papers add a fourth (place apple in drawer) to their average, so RoboFYI recomputes the three-task mean from the per-task values.
Simulation · Google Robot (RT-1 / Fractal) · SAPIEN (ManiSkill2)Report an error
89.0%Best · Xiaomi-Robotics-0
27Models ranked
Near saturationSaturation
Nov 18, 2026Numbers checked
| # | Model | Type | Score | Date | Source |
|---|---|---|---|---|---|
| 1 | VLA | 89.0%, self-reported by the authors, not a leaderboard entry | Feb 2026 | ||
| 2 | VLA | 88.3%, self-reported by the authors, not a leaderboard entry | Oct 2025 | ||
| 3 | VLA | 87.8%, self-reported by the authors, not a leaderboard entry | Aug 2025 | ||
| 4 | VLA | 85.8%, self-reported by the authors, not a leaderboard entry | May 2026 | ||
| 5 | VLA | 84.4%, self-reported by the authors, not a leaderboard entry | Aug 2025 | ||
| 6 | VLA | 82.7%, self-reported by the authors, not a leaderboard entry | Nov 2024 | ||
| 7 | VLA | 80.3%, self-reported by the authors, not a leaderboard entry | Dec 2024 | ||
| 8 | StarVLAStarVLA Team | VLA | 79.7%, self-reported by the authors, not a leaderboard entry | Apr 2026 | |
| 9 | open-pi-zeroopen-pi-zero (Allen Z. Ren) | VLA | 79.6%, self-reported by the authors, not a leaderboard entry | Jan 2, 2025 | |
| 10 | VLA | 78.7%, self-reported by the authors, not a leaderboard entry | May 2025 | ||
| 11 | VLA | 77.7%, self-reported by the authors, not a leaderboard entry | Jul 2025 | ||
| 12 | VLA | 73.8%, self-reported by the authors, not a leaderboard entry | Jan 2025 | ||
| 13 | VLA | 71.6%, self-reported by the authors, not a leaderboard entry | Aug 2025 | ||
| 14 | VLA | 70.6%, self-reported by the authors, not a leaderboard entry | Feb 2026 | ||
| 15 | Policy | 68.7%, self-reported by the authors, not a leaderboard entry | Mar 2025 | ||
| 16 | VLA | 67.6%, reported by a third party | May 2024 | ||
| 17 | VLA | 67.5%, reported by a third party | Jun 2026 | ||
| 18 | VLA | 63.0%, reported by a third party | Jul 2025 | ||
| 19 | VLA | 61.9%, reported by a third party | Aug 2025 | ||
| 20 | VLA | 60.5%, reported by a third party | May 2024 | ||
| 21 | VLA | 58.8%, reported by a third party | Aug 2025 | ||
| 22 | VLA | 57.9%, self-reported by the authors, not a leaderboard entry | Sep 26, 2025 | ||
| 23 | VLA | 57.4%, self-reported by the authors, not a leaderboard entry | Nov 2025 | ||
| 24 | VLA | 49.4%, reported by a third party | May 2024 | ||
| 25 | VLA | 46.2%, self-reported by the authors, not a leaderboard entry | Dec 2024 | ||
| 26 | VLA | 32.7%, reported by a third party | Jan 2025 | ||
| 27 | VLA | 14.6%, reported by a third party | May 2024 |
Metric: Visual-matching 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