XAM3500 is a small, secure and cost-effective module that supports many pre-trained AI models. At the heart of the module is K210 from Kendryte. A Dual Core 64bit RISC-V processor with FPU and dedicated CNN accelerator achieving 0.23TOPS under 300mW (module only). With the model embedded the module, a simple API call through one of the many interfaces makes adding AI capabilities to the system an easy task. With the IoT market in mind, security is a priority when designing the XAM3500. Utilizing the Microchip ATECC608A allows us to secure the module when communicating with cloud providers such as AWS and GCP.


  • K210: RISC-V Dual Core 64bit, with FPU
  • Convolutional Neural Network (CNN)
  • Hardware accelerator
    • 0.23 TOPS under 1W
    • Peripheral interfaces
    • GPIO
    • I2C
    • UART
    • SPI
    • DVP (camera)
    • LCD
  • On-chip Security
    • Supports Encrypted Firmware
    • 128-bit AES Accelerator
    • SHA256 Accelerator
  • Power Sources
    • 5V input
    • 3.3V output for IO Voltage
    • 1.8V output for IO Voltage
  • Dimensions
    • 33.02(W) x17.78(L) mm
  • Microchip ATECC608A
    • Trusted and Secure Authentication
      • Amazon AWS and Google GCP
    • Secure hardware-based key storage
    • Protected storage for up to 16 Keys, certificates or data
    • ECDH: FIPS SP800-56A Elliptic Curve
    • Diffie-Hellman
    • NIST standard P256 elliptic curve support
    • SHA-256 & HMAC hash
    • AES-128: encrypt/decrypt
  • Pre-trained model
    • Objects Detection (Customizable)
    • Face Detection
    • Age and Gender Estimation
    • Simple voice commands
    • Vibration abnormally detection


Product Brief


XaLogic GitHub
Kendryte K210 resources
Kendryte GitHub

Development Boards

Raspberry PI AI Hat – XAPIZ3500