The AIBOX 1684X features the SOPHON BM1684X AI processor, configurable with 16GB large memory and 128GB eMMC; it supports a peak computing power of 32 TOPS (INT8), 16 TFLOPS (FP16/BF16) computing power, and 2 TFLOPS (FP32) high-precision computing power; it supports 32-channel H.265/H.264 1080p@25fps video decoding; it supports mainstream programming frameworks, with a complete and user-friendly toolchain, and low algorithm migration cost; it is applicable to various AI computing scenarios such as visual computing, edge computing, general computing services, smart transportation, unmanned supermarkets, and drones.

01 Rich Algorithms, Comprehensive AI Private Deployment

Rich algorithms, #HotspotEngineProgram# comprehensive AI private deployment

02 Supports private deployment of ultra-large parameter models under Transformer architecture, such as large language models like LLaMa2, ChatGLM, Qwen, etc.

03 Equipped with the SOPHON BM1684X AI processor, featuring an octa-core ARM Cortex-A53 with a maximum frequency of 2.3GHz, built on a 12nm process; it possesses up to 32 TOPS (INT8) computing power, or 16 TFLOPS (FP16/BF16) computing power, or 2 TOPS (FP32) high-precision computing power, supports mainstream programming frameworks, and can be widely applied to AI inference in cloud and edge applications.

04 Powerful Multi-Channel Video AI Processing Performance Supports up to 32-channel 1080P H.264/H.265 video decoding, and 32-channel 1080P HD video full-process handling (decoding + AI analysis), meeting the demands of various AI application scenarios such as video stream facial detection and license plate recognition.

05 Supports dual 1000Mbps Ethernet, providing high-speed and stable network communication to meet the needs of different application scenarios.

Supports dual 1000Mbps Ethernet, providing high-speed and stable network communication to meet the needs of different application scenarios.