New agentic memory framework uses 118K tokens per query. LangMem burns through 3.26M.

NUS researchers' MRAgent framework reduces LLM agent memory retrieval to 118K tokens per query — vs. 3.26M for LangMem — using step-by-step reasoning.
Electronic Design

MoSys’ Graph Memory Engine Accelerator IP Targets Packet Classification on Intel’s Stratix 10 FPGAs

MoSys announced that its Graph Memory Engine (GME) Accelerator IP is now running on Intel Stratix 10 FPGAs. The GME is part of MoSys’ Packet Classification Platform, and is capable of performing ...

Microsoft unveils Memora to tackle AI agents’ memory problem

The research project promises more efficient long-term recall by organizing knowledge around abstractions and cue-based retrieval instead of raw conversation history.