M5Stack LLM Modul AX630C für large Language Model

The M5Stack LLM Module AX630C is a powerful, integrated module for offline processing of Large Language Models (LLMs). It provides an efficient and intelligent solution for applications such as smart home devices, voice assistants, or industrial controls. With its powerful AX630C SoC and 3.2 TOPs computing power at INT8, the module effortlessly supports complex AI tasks. Thanks to pre-installed models like Qwen2.5-0.5B, it enables natural interaction and operates completely independently of the cloud, ensuring data privacy and stability. With user-friendly integration via Arduino and UIFlow, you can easily incorporate the module into your existing projects.

Features:

• Powerful SoC: The AX630C processor offers up to 3.2 TOPs performance at INT8 for demanding AI tasks.
• Comprehensive memory support: With 4 GB LPDDR4 (1 GB for user applications) and 32 GB eMMC storage, the module is suitable for parallel models and multitasking.
• Offline usage: Enables AI interactions without an internet connection, ideal for privacy-sensitive applications.
• Versatile functionalities: Supports KWS (Wake Word), ASR (Automatic Speech Recognition), LLM (Large Language Model), and TTS (Text-to-Speech).
• Easy integration: Compatible with Arduino, UIFlow, and supports firmware upgrades via SD card or USB.
• Flexibility: Supports vision models like CLIP and YOLO and offers future updates for segmentation and depth detection.

Technical specifications:

• Manufacturer: M5Stack
• Manufacturer SKU: M140
• Processor: AX630C SoC with Dual Cortex A53 @1.2 GHz
• Computing power: Max. 12.8 TOPS @INT4, 3.2 TOPS @INT8
• Memory: 4 GB LPDDR4, 32 GB eMMC
• Audio: Microphone (MSM421A) and speaker (8Ω, 1W)
• Communication: Serial interface, USB-OTG
• Power consumption: 0.5 W in idle, 1.5 W at full load
• Operating temperature: 0–40°C
• Dimensions: 54 x 54 x 13 mm
• Weight: 49 g.