Powerful Computing Core · Efficient On-device AI Processing
[Strong AI Computing] 6TOPS＠INT8 NPU, multi-precision computing, image recognition, video analysis and intelligent detection without lag or frame drops
[Multi-core Computing] 8-core CPU (4xA76 + 4xA55), system operation, business logic and multi-task concurrency run rock-solid and smooth
Flexible Computing Expansion · On-demand Elastic Upgrade
[Professional Expansion Architecture] Independent cooling and power supply, 2 M.2 high-speed interfaces, no bandwidth bottleneck, stress-free expansion
[Multiple Computing Tiers] 20/60/160TOPS module options, up to 326TOPS heterogeneous computing, covering everything from small scale to ultra-high performance
Domain Controller 3588
Contact SalesChip Platform
RK3588
Main Processor
64-bit big/LITTLE core architecture, 4*Cortex-A76+4*Cortex-A55
Main Computing Power
6TOPS＠INT8
Expansion Computing Card
20TOPS/60TOPS/160TOPS INT8
GPU
ARM Mali-G610 MC4 GPU, dedicated 2D graphics acceleration module
ISP
32MP ISP, supports HDR and 3DNR
Memory
8GB LPDDR5, optional 16GB
Storage
128GB eMMC by default, optional 32G/64G
Expansion Storage
M.2 SSD: supports NVMe SSD 2280 (inside chassis)
Video Decoding
Decoding: H.265/H.264/AV1/VP9/AVS2, up to 8K＠60fps (7680X4320)
Power Supply
DC12V/5A
Cooling
Passive cooling
Operating Temperature
0°C~60°C
Operating Humidity
10%~90%RH, non-condensing
Dimensions
243mm×180mm×82mm
Protection Rating
IP51
Network Interface
RJ45*2: 10/100/1000Mbps network; WiFi: data forwarding WiFi/Bluetooth module
External Interface
USB3.0*4/SIM slot*1/RS-485*2/RS-232*1/DI*1/DIO*2/DEBUG*1/external power (5V*1)
Indicators
PWR*1/SYS*1/HDD status*1/WWAN (4G/5G status)*1
Expansion Interfaces
M.2(4G/5G)*1/M.2(SSD)*1/WIFI+BT*1/GPS & BeiDou dual-mode positioning
Model Performance · LLM
Model
Acceleration Chip
InputTokens
NewTokens
TTFT(mS)
TPOT(ms)
DecodeTPS
Qwen2.5-0.5B
RK182X
128
128
22.74
4.48
223.40
Qwen2.5-1.5B
RK182X
128
128
49.14
6.69
149.39
Qwen2.5-3B
RK182X
128
128
85.54
9.69
103.24
Qwen2.5-7B
RK1828
128
128
162.25
14.19
70.47
Qwen3-0.6B
RK182X
128
128
28.61
5.49
182.26
Qwen3-1.7B
RK1828
128
128
54.34
7.17
139.39
Qwen3-4B
RK1828
128
128
109.78
11.30
88.47
Qwen3-8B
RK1828
128
128
182.20
16.30
61.34
Model Performance · VLM
Model
Acceleration Chip
Vision Resolution
Vision(ms)
LLM TTFT(ms)
LLM DecodeTPS
FastVLM_1.5B_stage3
RK182X
512*512
168.85
49.83
151.01
InternVL3-2B
RK182X
448*448
184.19
49.85
147.62
InternVL3 5-4B
RK1828
448*448
176.99
110.06
87.95
Qwen2.5-VL-3B
RK182X
392*392
231.4
97.85
51.48
Qwen2.5-VL-3B
RK1828
392*392
212.28
87.6
104.05
Qwen2.5-VL-7B
RK1828
392*392
215.63
163.5
69.95
Qwen3-VL-2B
RK182X
384*384
114.38
56.55
142.00
Qwen3-VL-4B
RK1828
384*384
117.6
111.45
87.8
MiMo-VL-7B-RL
RK1828
392*392
216.56
173.59
64.97
MiniCPM_V_4
RK1828
448*448
236.67
97.81
106.56