M5Stack LLM630 Compute Kit powered by Axera Tech AX630C AI Soc with LLM and Computer Vision support for Edge AI

The M5Stack LLM630 Compute Kit is an advanced development tool built around the Axera Tech AX630C AI chip with a 3.2 TOPS NPU. It can handle tasks like computer vision and language models directly on the device, without needing to connect to the cloud.

The LLM630 Compute Kit comes with 4GB of LPDDR4 RAM and 32GB of eMMC flash storage. It offers both wired and wireless connectivity, featuring a JL2101-N040C Gigabit Ethernet chip and an ESP32-C6 module for 2.4GHz Wi-Fi 6. You can also connect a display and camera using the MIPI DSI and CSI connectors.

The kit comes with a 2.4GHz Wi-Fi antenna, a heatsink, and four EVA foam foot pads, making it ready for a variety of Edge AI applications and also comes with StackFlow AI framework that includes AI algorithms. It supports Jupyter Notebook for easy learning and development. 

Specifications of M5Stack LLM630 Compute Kit: 

SoC - Axera Tech AX630C

• CPU: 
• Dual-core Arm Cortex-A53 @ 1.2 GHz
• 32KB Instruction Cache (I-Cache)
• 32KB Data Cache (D-Cache)
• 256KB Level 2 Cache (L2-Cache)
• NPU:
• 12.8 TOPS (max) at INT4
• 3.2 TOPS at INT8
• ISP:  4K resolution at 30fps
• Video Processing:
• Encoding: 4K resolution
• Decoding: 1080p at 60fps (H.264 only)

Memory:

• 4GB LPDDR4
• 2GB system memory
• 2GB dedicated for hardware acceleration

Storage:

• 32GB eMMC 5.1 flash
• MicroSD card slot

Display Interface: 2-lane MIPI DSI connector (up to 1080p resolution at 30fps)

Camera Interface: 4-lane MIPI CSI connector (up to 4K resolution at 30fps, 1.25mm pitch)

Audio: 5-pin microphone and speaker header

Networking:

• Low-profile GbE RJ45 port (JL2101B-N040C)
• 2.4GHz WiFi 6 via ESP32-C6 SoC with SMA antenna

USB Interface: 1x USB 2.0 OTG Type-C port

Sensor: 6-axis BMI270 motion sensor

Expansion:

• 4-pin 2.0mm PORTA Grove connector (I2C)
• 4-pin 2.0mm PORTC Grove connector (UART)
• 8-pin 1.25mm pitch FUNC header (support system wake, power management, external LED control, I2C and more.)

Debugging/Programming: USB-C to UART via CH9102F USB to TTL chip

Miscellaneous:

• 2x buttons for power on/off
• user interaction, reset, and other functions

Power Supply:

• 5V/2A via USB-C port
• 1.25mm connector for 3.7V Li-ion/Li-polymer battery

Battery Management:

• AW32001ECSR charging chip
• BQ27220YZFR battery detection chip

Physical Specifications:

• Dimensions: 90.3 x 31.6 x 12mm
• Weight: 22.9 grams
• Temperature Range: 0°C to 40°C

The device operates on Ubuntu 22.04, and you can install AI models using .deb packages. For detailed technical information and installation instructions, please refer to the documentation website.The Axera AX630C AI SoC delivers performance comparable to the Jetson Orin Nano (in 10W power mode) for models such as Inception v1, MobileNet v1, and Yolov3s, but at a significantly lower cost.

It supports models like Qwen2.5-0.5/1.5B and Llama3.2-1B. The LLM630 Compute Kit is a compact, affordable alternative to options like the Raspberry Pi 5 with Hailo-8 AI accelerators or Rockchip RK3588 SBCs, depending on your project needs. Unlike the older M5Stack Module LLM, the LLM630 Compute Kit is a complete, standalone solution. 

The M5Stack LLM630 Compute Kit costs $69.99 and is available on AliExpress and the M5Stack store, temporarily out of stock because of the Chinese New Year holiday (January 21 to February 4) but will be back in stock after the holiday. It may also be available on M5Stack's Amazon store.