Module LLM (AX630C) Edge AI Development Board — Offline LLM Inference, 3.2 TOPs NPU, 4GB LPDDR4, 32GB eMMC

Module LLM is an all-in-one offline Large Language Model inference module built for edge devices that need responsive, private AI interactions. By running models locally with no cloud dependency, it preserves user privacy and delivers stable performance for applications from smart homes to industrial control. Typical uses include offline voice assistants, text-to-speech, smart home automation, interactive robots, and other embedded AI scenarios. Development is streamlined through integration with the StackFlow framework and UiFlow libraries, enabling developers to add intelligent features with only a few lines of code. The module is designed for straightforward integration into existing products without complex configuration, so manufacturers can quickly upgrade device intelligence. Onboard hardware supports natural voice interaction and data transfer: a built-in microphone and speaker, TF storage card slot, USB OTG, and an RGB status light cover common terminal needs. For expansion and maintenance, the onboard SD card slot allows cold/hot firmware upgrades, while the UART communication interface makes connections and debugging simple. The USB port provides master-slave auto-switching, functioning as a debugging interface and permitting attachment of additional USB peripherals such as cameras. The module is plug-and-play with M5 hosts for immediate AI interaction. Performance is driven by the AX630C SoC, which integrates a high-efficiency 3.2 TOPs NPU with native support for Transformer models to handle complex inference tasks. Memory and storage — 4GB LPDDR4 and 32GB eMMC — enable parallel model loading and sequential inference of multiple models, ensuring smooth multitasking and reliable edge AI operation.