SO-101 Bench leaderboard

Language-conditioned spatial manipulation benchmark for a single SO-101 arm, run in an Isaac Lab twin with a real-robot rank-transfer check. All learned policies share the same 3,629 sim demos and training budget; the authors run no public submission process.

Low-cost arm · SO-101 · Isaac Lab (digital twin)Report an error
83.2%Best · GPT-6.1 Sol (PyRUA) + Galaxea G0.5
11Models ranked
OpenSaturation
Nov 18, 2026Numbers checked
#ModelTypeScoreDateSource
1GPT-6.1 Sol (PyRUA) + Galaxea G0.5OpenAI model; SO-101 Bench harnessAgent system83.2%, self-reported by the authors, not a leaderboard entryOct 27, 2026
2DeepSeek V4.1 Flash (PyRUA) + Galaxea G0.5DeepSeek model; SO-101 Bench harnessAgent system66.8%, self-reported by the authors, not a leaderboard entryOct 27, 2026
3InternW0-ΔShanghai AI LabWorld-action61.4%, self-reported by the authors, not a leaderboard entryOct 27, 2026
4Galaxea G0.5Galaxea AIVLA60.8%, self-reported by the authors, not a leaderboard entryOct 27, 2026
5GR00T N1.7NVIDIAVLA45.8%, self-reported by the authors, not a leaderboard entryOct 27, 2026
6ImageWAMShanghai Jiao Tong UniversityWorld-action43.0%, self-reported by the authors, not a leaderboard entryOct 27, 2026
7π0.5Physical IntelligenceVLA41.0%, self-reported by the authors, not a leaderboard entryOct 27, 2026
8LaWAMTsinghua UniversityWorld-action39.9%, self-reported by the authors, not a leaderboard entryOct 27, 2026
9MolmoAct2Ai2VLA23.8%, self-reported by the authors, not a leaderboard entryOct 27, 2026
10MolmoAct2-SO101Ai2VLA20.9%, self-reported by the authors, not a leaderboard entryOct 27, 2026
11FLUX 3 ActionBlack Forest LabsWorld-action8.0%, self-reported by the authors, not a leaderboard entryOct 27, 2026

Metric: Mean success over 15 task-family × object-split cells (916 episodes, 95% bootstrap CI). 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