AgiBot's GO-3 Nano Preview tops RoboDojo benchmark with 61.90 average score

A preview of AgiBot's robot manipulation foundation model GO-3 Nano Preview has taken first place on RoboDojo's simulation leaderboard with an average score of 61.90 and a 56.42% success rate. Second place PyRUA-Lean scored 55.26, a 6.64-point gap. The model leads in precision manipulation, long-horizon execution and open-ended semantics but trails in memory and generalisation.

• Average score 61.90 and 56.42% success rate on RoboDojo simulation board
• 6.64-point lead over runner-up PyRUA-Lean at 55.26
• Leads precision manipulation (78.65) and long-horizon execution (68.21)
• Weak on randomised generalisation with 26.97 points