This product qualifies for Free Surface Shipping !
B2B GST Credit Available
- Stock: 0 in Stock
- SKU: 08439
- Delivery Time
- Bulk & B2B RFQ
Rockchip LQ50 M.2 2280 AI Accelerator Module – 160 TOPS INT8 Edge AI Card for Local LLM & VLM Inference
The Rockchip LQ50 M.2 2280 AI Accelerator Module is a high-performance edge AI accelerator designed to bring powerful on-device artificial intelligence to PCs, industrial systems, embedded platforms, and edge servers. Delivering up to 160 TOPS (INT8) AI computing performance, the LQ50 enables fast and energy-efficient execution of Large Language Models (LLMs) and Vision Language Models (VLMs) without relying on cloud processing.
Built in the standard M.2 2280 M-Key form factor with a PCIe Gen4 x4 interface, the module integrates easily into compatible systems, making it ideal for AI-powered edge computing, robotics, industrial automation, intelligent video analytics, autonomous machines, and private AI deployments. Its high-bandwidth LPDDR5 memory architecture and low power consumption allow real-time inference while maintaining excellent energy efficiency. The module supports major operating systems including Windows 11, Linux, and Android, making development and deployment straightforward across multiple platforms.
Designed for local AI inference, the LQ50 significantly reduces latency, enhances data privacy, and minimizes cloud dependency. It is well-suited for running modern AI models such as Llama, Qwen, ChatGLM, and multimodal vision-language applications directly at the edge, providing responsive AI experiences for industrial and commercial environments.
Features
- Up to 160 TOPS INT8 AI computing performance
- Standard M.2 2280 (22 × 80 mm) M-Key form factor
- PCIe Gen4 x4 high-speed host interface
- Optimized for Local LLM and VLM inference
- Supports edge AI, generative AI, computer vision, and multimodal AI
- High-bandwidth LPDDR5 memory architecture
- Low power consumption for energy-efficient AI deployment
- Compatible with Windows 11, Linux, and Android
- Ideal for industrial AI, robotics, smart cameras, edge servers, and AI workstations
- Reduces cloud dependency while improving privacy and inference latency
Specifications
- Form Factor: M.2 2280 (22 × 80 mm)
- Interface: PCIe Gen4 x4, M-Key
- AI Performance: 160 TOPS @ INT8
- FP Performance: Up to 100 TFLOPS @ bFP16
- Memory: LPDDR5 (configuration dependent)
- Memory Bandwidth: Up to 153.6 GB/s
- Operating Systems: Windows 11, Linux, Android
- Typical Power Consumption: ≤13W
- AI Workloads: LLM, VLM, Computer Vision, Edge AI
- Cooling: Supports passive or active cooling solutions
Applications
- Local Large Language Model (LLM) inference
- Vision Language Model (VLM) applications
- AI assistants and chatbots
- Industrial machine vision
- Smart surveillance and video analytics
- Robotics and autonomous systems
- Edge AI servers
- Intelligent transportation systems
- Healthcare AI devices
- Retail analytics and smart manufacturing
Package Includes
- 1x Rockchip LQ50 M.2 2280 AI Accelerator Module – 160 TOPS INT8 Edge AI Card for Local LLM & VLM Inference
Warranty Terms :
Branded products (Waveshare/ Raspberry Pi/ Radxa/ DWIN etc) are as per brand warranty terms. Generic products are eligible for replacement within 7 days of delivery based on following conditions.
Products are eligible for a replacement if there is physical damage during transit or a clear manufacturing defect. Any claim for damage/ wrong items/ replacement must be made within 48 hours of successful delivery of the product. Requests for refund of amount or exchange of product(s) will be reviewed on a case by case basis and HUBTRONICS carries final rights to take a final decision on refund, store credit or exchange.
What to take care of ?
Most of our products are technical in nature and require trained personnel to handle them. Any kind of mishandling can result in damage or malfunctioning of the product, it is not possible for us to provide a replacement or refund if the product is damaged due to technical or physical mishandling.
Who pays the return shipping cost?
Free of cost return pickup from delivery address is eligible for products categorized as manufacturing defect/ Damage in transit/ Wrong items delivered. We do not support return pickup for any B2B order or offline orders not placed on website. Buyer need to arrange return shipping for B2B orders.
What voids warranty?
Please note that replacement/ return or refund is not possible if the product is not compatible with your system/application. Please check all the technical details/ dimensions etc. before ordering and contact us for any kind of clarifications.
If product is soldered/ programmed at any point or physically altered in any way, it is not eligible for return/ replacement. If the item requires soldering for testing, please ensure that you use suitable clips or jig to test it without soldering. Any kind of soldering job makes the product ineligible for return or replacement.
Following items are not eligible for return under any circumstances
- Device Programmers / Dongles
- Soldering Consumables (Wire, Flux etc)
