Tag
ActiveMem introduces a distributed active memory system that decouples agent memory from the core LLM reasoning process, achieving state-of-the-art accuracy on long-horizon tasks with significantly reduced overhead.