Firefly Qualcomm Product Portfolio: Covering Full-Chain Requirement Scenarios
As edge computing and on-device AI accelerate their penetration, embedded hardware is rapidly evolving toward “tiered computing power, diverse form factors, and precise scenario targeting.” As a leading embedded solution provider in the industry, Firefly has built a comprehensive product portfolio based on Qualcomm’s three core platforms, covering the full spectrum of needs from entry-level prototyping to industrial-grade high-compute deployment.
QCS6490 Series
12 TOPS Compute Power: The Entry-Level AI Powerhouse
As a representative platform in Qualcomm’s Q6 series, the QCS6490 is built on an advanced 6nm process. Firefly has launched a compact SBC (Single Board Computer) based on the QCS6490, achieving a balanced combination of computing power and connectivity. It is the ideal choice for prototyping, smart terminals, and lightweight edge AI.
Octa-Core AI Processor QCS6490
The QCS6490 features an 8-core Kryo CPU (1+3+4 tri-cluster architecture with a max frequency of 2.7GHz), integrated Adreno 643 GPU, and the 6th Gen Qualcomm AI Engine. With an NPU delivering 12 TOPS, it smoothly runs YOLO-series detection models for real-time object recognition and visual analysis. Its multimedia capabilities are equally impressive, supporting up to 5 concurrent camera inputs, 4K@60fps video decoding, and 4K@30fps encoding, meeting the demands of multi-camera vision and edge gateways.
Robust Network Connectivity
In terms of network connectivity, the QCS6490 series supports Gigabit Ethernet, Wi-Fi, and Bluetooth, along with expansion interfaces like USB 3.1 and PCIe Gen 3 for flexible adaptation to various peripherals and sensors. On the software side, it offers long-term support for Android, Linux, Ubuntu, and Windows 11 IoT Enterprise, lowering the barrier to cross-platform development.
Lightweight Edge AI Application Scenarios
Widely applicable to smart retail terminals, industrial handheld devices, lightweight AI vision gateways, educational development kits, and human-machine interfaces (HMI).
QCS8550 Series
48 TOPS Compute Power: Dual Form Factors for Consumer and Industrial Scenarios
Based on Qualcomm’s flagship Q8 series platform, the QCS8550 product line serves as the primary compute powerhouse in Firefly’s Qualcomm lineup. It offers two hardware form factors—the AIBOX smart box and the EC industrial computer—catering to commercial AI deployment and harsh industrial environments respectively.
High-Performance AI Processor QCS8550
Built on a 4nm process, the QCS8550 features an hexa-core Kryo CPU architecture (1 Prime core at 3.36GHz + 2 performance cores + 3 efficiency cores). Its NPU compute power surges to 48 TOPS, supporting INT4 / INT8 / INT16 / FP16 mixed-precision operations with significantly improved energy efficiency over previous generations. Its biggest highlight is the support for private, on-device deployment of large language models (LLMs). It natively runs mainstream Transformer-based models like Qwen, Llama, Gemma, and DeepSeek, empowering edge intelligence with natural language understanding and multimodal interaction capabilities.
Graphics and Multimedia Codec Capabilities
The Adreno 740 GPU supports hardware ray tracing and is compatible with graphics APIs such as Vulkan 1.2 and OpenCL 3.0. Video encoding/decoding supports 8K@60fps decoding and 8K@30fps encoding, capable of processing multiple high-definition video streams simultaneously.
Dual Coverage for Consumer and Industrial Scenarios
Widely applicable to edge AI inference servers, smart security NVRs, industrial vision inspection, autonomous mobile robots (AMR), video conferencing terminals, and cloud gaming edge nodes.
IQ-9075 Series
200 TOPS Industrial-Grade Compute: High-Reliability AI Platform
Targeting the most demanding industrial edge AI scenarios, Firefly introduces the IQ-9075 AIBOX based on Qualcomm’s IQ9 series. Topping the embedded AI performance tier with 200 TOPS of INT8 dense compute, it is purpose-built for high-load, long-lifecycle, and extreme-environment industrial deployments.
Flagship AI Processor IQ-9075
The IQ-9075 is equipped with an octa-core Kryo Gen 6 high-performance CPU and an Adreno 663 GPU. The Hexagon Tensor Processor integrates vector and matrix acceleration units, enabling the native operation of 13-billion-parameter LLMs at an inference speed of 12 tokens/s, achieving true on-device generative AI capabilities.
High-Performance 8K VPU Codec
The AIBOX-9075 supports concurrent access to 8 GMSL cameras and can perform real-time AI analysis on 4K video streams from 4 channels simultaneously. Paired with the Spectra 690 ISP and multiple DSPs, it effortlessly handles complex vision tasks such as multi-camera surround view, industrial quality inspection, and behavior analysis. It also supports CAN, RS485, and 4G/5G expansion.
Industrial-Grade High Reliability
Industrial-grade reliability is the core trait of this platform: it supports an ultra-wide operating junction temperature range of -40°C to 115°C, features ECC-enabled LPDDR5 memory (up to 36GB), and is covered under Qualcomm’s long-term product lifecycle program. The software system is built on Linux Yocto and Ubuntu, integrating a real-time subsystem and security encryption engines to meet industrial functional safety and data compliance requirements.
Industrial-Grade AI Application Scenarios
Widely applicable to industrial AI quality inspection equipment, high-end robot controllers, multi-camera smart analysis terminals, medical imaging equipment, industrial edge gateways, and extreme-environment embedded computing.
Firefly Qualcomm Product Portfolio
From Raspberry Pi-sized lightweight SBCs to dual-form-factor 48 TOPS flagships, and up to the 200 TOPS industrial-grade powerhouse, Firefly’s Qualcomm product line has formed a clear compute gradient and form factor matrix:
Whether for AI algorithm validation, smart terminal development, production line implementation, or large-scale deployment, developers can find a precisely matched hardware foundation within Firefly’s Qualcomm product portfolio. Backed by Qualcomm’s mature software ecosystem and Firefly’s comprehensive technical support system, developers can significantly shorten the product cycle from prototype to mass production, accelerating the deployment of edge AI solutions.
———————————————
Access Firefly Store to know more