Top-Tier AI Compute
48 TOPS INT8 / 12 TOPS FP16 easily runs LLMs up to 14B parameters.
Rhino Core Q85
A mainstream on-device AI System-on-Module powered by Qualcomm QCS8550, with 48 TOPS INT8 / 12 TOPS FP16 in a 40×40mm footprint for robotics, edge AI and high-end smart devices.
QCS855048 TOPS INT840×40mm13WWi-Fi 7Ubuntu22.04
Highlights
Professional 8K Imaging
8K@60 decoding and 8K@30 recording with AV1 support, plus up to 8 MIPI CSI lanes.
Zero-Code Model Adaptation
AIMO toolchain, AidLux OS and 500+ pretrained models on ModelFarm.
Fusion OS
AidLux runs Android and Ubuntu side by side with seamless switching, making development simple.
Specifications
Chipset process
Qualcomm QCS8550, 4nm process
CPU
1 x Gold Plus @ 3.2GHz + 2 x Gold @ 2.8GHz + 3 x Silver @ 2.0GHz
GPU
Adreno 740
Performance: 4.4 TFLOPS (FP16), with ray tracing support
AI compute
48 TOPS (INT8) / 12 TOPS (FP16)
Core: Hexagon tensor processor with 4× HVX vector coprocessors & 1× HMX matrix coprocessor
Memory (LPDDR5X)
12GB / 16GB LPDDR5X
Storage
128GB UFS 3.1
Operating system
Ubuntu 22.04
Android 14
AidLux Fusion OS 3.0(Android+Ubuntu)
Video processing (VPU)
Decode: 8K@60fps / 4K@240fps H.264/H.265/VP9/AV1
Encode: 8K@30fps / 4K@120fps H.264/H.265
Camera interface
5× MIPI CSI D-PHY: 4× 4-lane + 1× 2-lane
PCIe
1× PCIe 2-lane Gen4 + 1× PCIe 2-lane Gen3
USB
1× USB 3.1 Gen 2 Type-C (supports DP 1.4)
Wireless & network
Wired: 2× Gigabit RJ45
Wireless: Wi-Fi 7 / Bluetooth 5.3
General interfaces
Serial: 8 ports (2 configurable as RS485)
CAN bus: 2
GPIO: 10 (1.8V level)
Other: I2C, SPI, I2S and more
Power input
3.5~4.2V, 3.8V typical
Power consumption
CPU/GPU/NPU typical full-load power: 13W
Package & mounting
40×40mm
BTB board-to-board connector
Product lifecycle
10 years
Build with Rhino Core Q85
Get Rhino Core Q85 documentation, evaluation kits, pricing and support for your Physical AI project.