Agent Memory Leaderboard launches a second public evaluation challenge for agent memory systems across textual, coding, and multimodal tracks. Participants deploy their own Add/Search APIs while the platform handles evaluation orchestration, with 150,000 RMB in prizes for open-source submissions and public rankings by track.