Radxa Adds Snapdragon and Dragonwing rCore AI Modules
Radxa's product catalog now lists two new Qualcomm-based systems-on-module in its rCore lineup: the Snapdragon 8cx Gen 3-powered rCore-Q8280 and Dragonwing QCS6490-powered rCore-Q6490. Radxa rates them at 29+ TOPS and 12 dense TOPS, respectively, for embedded edge-AI workloads, but both official product pages still list availability as coming soon.
Radxa's product catalog now lists two Qualcomm-based systems-on-module in its rCore family. The larger rCore-Q8280 uses Qualcomm's Snapdragon 8cx Gen 3, while the compact rCore-Q6490 uses the Dragonwing QCS6490. LinuxGizmos independently reported the pair on August 10 as an expansion of Radxa's embedded-AI module lineup.
Two modules for different integration targets
Radxa rates the rCore-Q8280 at 29+ TOPS through Qualcomm's AI Engine. Its official specifications list an eight-core Kryo CPU, Adreno 690 graphics, as much as 32 GB of LPDDR4X memory, UFS and NVMe storage options, multiple display interfaces, and a 50 x 80 mm BGA package. Radxa positions it for edge AI, multimedia processing, smart terminals, and industrial systems.
The rCore-Q6490 is smaller at 42 x 36 mm. Radxa says its Dragonwing QCS6490 combines an eight-core Kryo 670 CPU, Adreno 643 GPU, and Hexagon 770 AI engine rated at 12 dense TOPS. The module supports up to 16 GB of LPDDR5, five four-lane MIPI CSI camera interfaces, PCIe expansion, and eMMC, UFS, microSD, or NVMe storage. Those interfaces make the lower-compute module more directly oriented toward camera-heavy robotics and machine-vision designs.
| Module | Qualcomm platform | Vendor-rated AI performance | Memory ceiling | Form factor |
|---|---|---|---|---|
| rCore-Q8280 | Snapdragon 8cx Gen 3 | 29+ TOPS | 32 GB LPDDR4X | 50 x 80 mm BGA |
| rCore-Q6490 | Dragonwing QCS6490 | 12 dense TOPS | 16 GB LPDDR5 | 42 x 36 mm BGA |
What developers should verify
The TOPS figures are vendor specifications, not independent application benchmarks, so they do not establish model latency, power draw, or sustained throughput in a deployed system. Radxa's pages list operating-system and storage support, but both sales sections currently say coming soon. Teams evaluating the modules should therefore treat availability, pricing, carrier-board support, and production software maturity as unresolved until Radxa publishes orderable configurations and deployment documentation.
Key Points
- 1The rCore-Q8280 pairs Snapdragon 8cx Gen 3 with a vendor-rated 29+ TOPS and up to 32 GB of LPDDR4X.
- 2The smaller rCore-Q6490 uses Dragonwing QCS6490, offers a vendor-rated 12 dense TOPS, and exposes five four-lane MIPI CSI camera interfaces.
- 3Both official product pages currently say coming soon, so pricing and orderable configurations remain unannounced.
Scoring Rationale
A relevant embedded edge-AI hardware expansion with concrete compute, memory, and camera-interface specifications, moderated because both modules remain listed as coming soon and vendor TOPS claims lack independent workload benchmarks.
Sources
Primary source and supporting public references used for this report.
Practice with real Ad Tech data
90 SQL & Python problems · 15 industry datasets
250 free problems · No credit card
See all Ad Tech problems
