Yuanli Lingji's embodied AI model DM0.5 has topped the RoboDojo benchmark with a comprehensive score of 24.90 and an average success rate of 19.34%, excelling particularly in memory tasks. The model is open-sourced and has achieved leading results across multiple benchmarks, demonstrating strong generalization and fine manipulation capabilities.
SynthePulse Insight · AI deep readingMembers
Version 1 · 1 source
Yuanli Lingji DM0.5 tops the RoboDojo benchmark with a composite score of 24.90 and an average success rate of 19.34%, and is fully open-sourced. Its long-term memory capability is a key advantage, and it also performs excellently across multiple other benchmarks.
RoboDojo, initiated by the Multimedia Laboratory at the University of Hong Kong, together with UC Berkeley, Tsinghua University, and nearly 20 other top academic institutions worldwide, is known for its high difficulty and realism. As of July 2026, the average success rate of top models on the simulation leaderboard is only 8.80%, and on the real-robot leaderboard only 12.8%. DM0.5 tops the charts with a composite score of 24.90 and an average success rate of 19.34%, becoming the new SOTA.
DM0.5's standout strength is its memory capability. In the Cover Blocks task, where the robot must first cover randomly arranged red, green, and blue blocks and then uncover them in order, DM0.5 achieved a 100% success rate across three random tests. On the Memory dimension of the RoboDojo leaderboard, DM0.5 scored 47.74, significantly ahead, which was key to its top ranking.
The information in this article primarily comes from a report by QbitAI, which is a secondary source. All data is derived from that report and has not been independently verified by the first party. Some performance figures (such as RoboDojo scores and success rates on other benchmarks) are as claimed by the source and may be influenced by evaluation conditions.
Primary report
Primary source