NXP unveils edge AI board as open-source Zero-Mem framework boosts AI memory efficiency

NXP Semiconductors introduced the FRDM i.MX 95 Pro development board, built around the i.MX 95 application processor with six Arm Cortex‑A55 cores, Cortex‑M33 and Cortex‑M7 real‑time cores, and the integrated eIQ Neutron Neural Processing Unit (NPU). The board offers 16 GB LPDDR5 memory, 32 GB eMMC storage, Wi‑Fi 6 tri‑radio connectivity, dual M.2 slots and multiple camera/display interfaces, targeting edge AI, robotics, machine vision and smart automation applications.

Researchers released Zero‑Mem, an open‑source long‑term memory framework for AI agents that eliminates token consumption during memory storage and retrieval. By using an entity‑context graph combined with a temporal hierarchy, Zero‑Mem stores original interaction traces and only calls a large language model (LLM) at the final answer stage, cutting inference costs and latency by about 57.6 % in benchmarks while preserving accuracy. The framework aims to improve scalability for edge and enterprise AI deployments.