Multi-device Task Orchestration on HeraBench benchmark leaderboard · SOTA2 Research

Multi-device Task Orchestration on HeraBench

75.84Completion RateH-RePlan

Evaluation Results

Method	Links
H-RePlan2026.06		75.84	77.72	36.78	710,871	1,932,765
UFO³2026.06		61.05	67.81	0	321,802	—
UFO³2026.06		46.86	56.67	13.79	1,449,681	10,512,553
CRAB2026.06		28.84	42.8	0	488,579	—
CRAB2026.06		2.16	9.3	0	547,342	—