RoboChallenge Table30 V2 leaderboard

Centralized remote evaluation service run by Dexmal, where submitted policies execute on the organizer’s real robots. V2 results are tracked separately from the original 30-task v1 board.

Real robot · Multiple (UR5, Franka, ARX5, Cobot Magic ALOHA in v1)Report an error
43.3%Best · LoopWAM
12Models ranked
OpenSaturation
Nov 18, 2026Numbers checked
#ModelTypeScoreDateSource
1LoopWAMLoopWAM TeamWorld-action43.3%Sep 24, 2026
2VLA-DM0.5KDDI ResearchVLA40.7%Sep 16, 2026
3ambaseAVerse001VLA33.0%Nov 12, 2026
4my16TymtboVLA30.7%May 15, 2026
5JV0JIIOVVLA16.0%May 11, 2026
6mcmcBrainsVLA15.3%May 15, 2026
7MagicBotMagicBotVLA14.7%May 14, 2026
8holobrainHRLVLA14.3%May 14, 2026
8π0.5Physical IntelligenceVLA14.3%, second-hand number, not confirmed at the sourceApr 15, 2026
10pi05_loomsmartVLAVLA12.3%Oct 27, 2026
11W0WMAVLA12.0%May 14, 2026
12hcp-eHCP-EVLA10.3%May 8, 2026

Metric: Task success rate on the V2 real-robot task set. 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