Hardware Sipeed Crams 32GB LPDDR5 and a 60 TOPS NPU Into a Compact RISC-V Board That Hits 15 Tokens/s on Qwen-3.5 35B AI LLMs Hassan Mujtaba • at EDT Add on Google Sipeed has unveiled its new RISC-V powered SBC platform, the K3 series, which can run up to Qwen3.5 35B AI LLMs and start at $299. The Sipeed K3 SBCs Pack A Punch For AI LLMs In a Compact "Pico-ITX" Form Factor, Up To 32 GB Memory Support Chinese manufacturer, Sipeed, has partnered with SPACEMIT to launch its brand new K3 series SBCs or Single-Board Computers (such as the Raspberry Pi). These PCs are normally aimed at AI, Edge, and Networking applications. Sipeed has leveraged SPACEMIT's Key Stone K3 AI CPUs, which are based on the RISC-V architecture, and offer some decent AI LLM capabilities for their size. Related Story This PCIe AI Accelerator Card Can Run 700B LLMs Locally With 384 GB Memory at Just 240W, Less Than Half The Power of RTX PRO 6000 BlackwellSome highlights of the K3 series include: Brutal Power - 8x X100 High-Performance Cores + 8x A100 AI Cores. Fusion architecture delivering 130k DMIPS general computing power. LLM Native - Run 30B parameter models locally and smoothly. Inference speed > 10 Tokens/s. No cloud needed, private intelligence. Blazing Speed - 32GB LPDDR5 Memory (6400Mbps) with 51GB/s bandwidth. Equipped with full-speed PCIe Gen3 and USB 3.0. Seamless Upgrade - Hardware compatible with Jetson Orin Nano carrier boards. Reuse existing cases and ecosystems, migrating from ARM to RISC-V smoothly. Starting with the specifications, the SPACEMIT K3 AI CPU comes with 8 X100 high-performance cores and 8 A100 AI cores. These are part of the fusion architecture, which delivers up to 130,000 DMIPS for general-purpose computing. The chip operates at a clock speed of 2.4 GHz and is equivalent to the ARM-A76 CPU in terms of performance. For the AI side, the K3 AI NPU offers up to 60 TOPS of AI compute capabilities, supporting various data types such as BF16, FP16, FP8, INT8, and INT4. The SBCs are rated to easily run up to 30B AI LLMs & have scored a 84% intelligence rating of a 235B model. As a demonstration, Sipeed states that the K3 can run Qwen-3.5 35B easily, delivering up to 15 Tokens/s. In terms of memory support, the CPU comes with LPDDR4X-4266 and LPDDR5-6400 support in up to 32 GB capacities. The total memory bandwidth is rated at up to 51 GB/s for the LPDDR5 solution. Sipeed is listing three models of the K3 in 8 GB, 16 GB, and 32 GB flavors. Coming to the design, the K3 CoM260 Kit is a 69.6mm x 45mm module that features a 260-pin SO-DIMM slot and is compatible with NVIDIA's Jetson Orin series. The platform supports Ubuntu 26.04 & ROS, with a rich set of IO capabilities. As for the K3 Pico-ITX, the board comes in a 100mm x 86mm form factor, and has rich IO support, such as: 10GbE Strike: Onboard 10 Gigabit Ethernet + Gigabit LAN. No more I/O bottlenecks for edge computing. Modern Power: Dual Type-C ports with USB-PD + Alt-DP support. Say goodbye to bulky bricks; power and debug with a single cable. Unified Memory: Onboard 16GB/32GB LPDDR5 unified memory, high bandwidth meets AI model throughput demands. The full specifications of the Sipeed K3 series are listed below: Hardcore Specs SoCSpacemiT Key Stone K3 (8x X100 + 8x A100 RISC-V Cores)AI Performance60 TOPS (INT4), Supports BF16/FP16/INT4Memory (RAM)8/16/32GB LPDDR5 @ 6400MT/s (51GB/s Bandwidth)StorageSupports eMMC 5.1, SD Card, M.2 NVMe SSD (PCIe Gen3 x4)MultimediaDecode: 4K@120fps (H.265/VP9) | Encode: 4K@60fpsOS SupportBianbu OS (Debian based), Docker Support, RISC-V KVM VirtualizationDimensions103mm x 90.5mm x 35mm (with heatsink) There's a $10 US difference between the Kit and the ITX board, but you are looking at a starting price of $299-$309 for the 8 GB SBCs and up to $629-$639 for the 32 GB configurations. Currently, NVIDIA's Jetson Orin Nano with 8 GB memory and a 6-core ARM CPU retails for $247.99. If you want a small AI machine that can handle up to 30B LLMs, then these definitely look enticing. About the : A Software Engineer by training and a PC enthusiast by passion, Hassan Mujtaba serves as 's for hardware section. With years of experience in the industry, he specializes in deep-dive technical analysis of next-generation CPU and GPU architectures, motherboards, and cooling solutions. His work involves not only breaking news on upcoming technologies but also extensive hands-on reviews and benchmarking. Follow on Google to get more of our news coverage in your feeds. Further Reading Mobile DRAM Prices Expected To Increase By ~100% Quarter-Over-Quarter, As Long-Term Agreements Now Getting Signed At Prices As High As $21/GB DDR6 Memory Development Kicks off As Samsung, SK Hynix & Micron Race Towards Commercialization By 2028-2029 Apple Quietly Scoops Up Any Available LPDDR5 Supply While Freezing iPhone Prices, Pushing Chinese OEMs To Kill Their Ultra Flagships Agentic AI Pushes CPUs to Pack 400 GB of Memory, 4x More Than Today, as DRAM Shortage Spirals Toward 2027 Read all on Sipeed Crams 32GB LPDDR5 and a 60 TOPS NPU Into a Compact RISC-V Board That Hits 15 Tokens/s on Qwen-3.5 35B AI LLMs
