SE3 top image for mobile

AI mini SE3

The SOPHON AI mini SE3 is a deep learning-based edge computing box. It is equipped with the TPU chip BM1682 independently developed by Bitmain. The floating point 32-bit peak computing power is up to 3TFLOPS. It can process 4 channels of HD video at the same time.With diversified algorithms, it can realize face comparison, control analysis, video structure, object recognition, etc.

Extraordinary design, born for efficiency

Real-time response

It can be connected to different types of collection devices to calculate the nearest data to achieve real-time response and edge control

Open system

Deploy various intelligent algorithms and applications of customers and partners to achieve rapid AI empowerment for various industry application scenarios

Edge Cloud Collaboration

The "edge-cloud" collaborative distributed architecture distributes the computing load between the edge and the cloud, which can not only achieve scalable and ultra-large-scale management, but also have the characteristics of real-time response and control of the edge

High-precision computing power


Fast identification

Support up to 50,000 face database, 0.5 second for each recognition

High performance processing power

Support 4 video streams recognition or 10 image stream recognition

Rich algorithms

Support various algorithms such as person / vehicle / non-vehicle/ object recognition, video structuring, trajectory analysis, etc.

Rich scenes

Support smart Park / smart security / industrial application / business application and other fields and scenarios for flexible deployment

Quick transplant

Support Caffe / TensorFlow / PyTorch / MXNet / Paddle Lite and other mainstream deep learning frameworks

Wide application and rich scenes

Support multiple intelligent algorithms such as face recognition, vehicle recognition, object recognition, video structure, behavior analysis, etc

Easy to use and convenient, full stack efficient

BMNNSDK (BITMAIN Neural Network SDK) one-stop toolkit provides a series of software tools such as the underlying driver environment, compiler, inference deployment tool and so on. Easy to use and convenient, covering the model optimization, efficient runtime support and other capabilities required for the neural network inference stage, providing easy-to-use and efficient full-stack solutions for deep learning application development and deployment. BMNNSDK minimizes the development cycle and cost of algorithms and software. Users can quickly deploy deep learning algorithms on various AI hardware products of Fortune Group to facilitate intelligent applications.

Support mainstream programming framework







4-core A53@1.8GHz

Computing power


Peak computing power 3TFLOPS

Memory and storage





External interface

Ethernet interface

10/100/1000Mbps adaptive


MicroSD *1


Length * width * height

210mm * 115mm * 45mm

Power supply and power consumption

Power supply

DC 12V

Typical power consumption

≤35W (depending on the configuration)

Temperature and humidity

Working temperature

-10℃ ~ +45℃


10% ~ 90%, no condensation