SHANGHAI, Oct. 22, 2026 /PRNewswire/ -- The LattePanda Team today launched LattePanda Sigma 2, an x86 single-board server designed to bring high-performance local AI computing directly into intelligent products and systems. Built around the Intel^® Core™ Ultra X7 358H processor with Intel Arc™ B390 graphics, LattePanda Sigma 2 delivers up to 180 TOPS of total AI performance at INT8 in a compact form factor.

As "The Embeddable AI Single Board Server," LattePanda Sigma 2 combines high-performance computing, up to 64GB of memory, high-speed networking, flexible expansion, and real-time hardware control in a single board. Rather than sitting alongside a product, it is designed to be integrated directly into the product itself, supporting applications from autonomous robots and vision systems to professional instruments and private AI appliances.

180 TOPS of AI Performance on a Single Board

LattePanda Sigma 2 combines the Intel^® Core™ Ultra X7 358H CPU, Intel Arc™ B390 GPU, and integrated NPU to deliver up to 180 TOPS of total AI performance at INT8. It features a 16-core CPU with up to 4.8 GHz clock speed, a 12 Xe-core Intel Arc B390 GPU delivering up to 122 TOPS of INT8 AI performance, and an NPU capable of up to 50 TOPS.

With up to 64GB of LPDDR5X-8533 RAM and 136.5 GB/s memory bandwidth, LattePanda Sigma 2 provides the capacity and throughput required for demanding local AI workloads. Up to 59GB of VRAM can be allocated for AI workloads, providing enough room for 35B-class quantized MoE models and the KV cache required by long-context AI agents. With INT4 quantization and OpenVINO™, the system can run the Qwen3.6-35B-A3B model locally at around 30 tokens per second.

Built for Local AI Systems

LattePanda Sigma 2 is designed to support both individual AI workloads and multi-agent local AI systems. Developers can expand compute capacity through M.2, OCuLink, or Thunderbolt™ 4, connecting additional AI accelerators as workloads grow. An integrated RP2350 microcontroller adds GPIO connectivity and real-time hardware control, allowing AI applications to interact with sensors and actuators. Preloaded MicroPython firmware provides a programmable interface for intelligent devices and automation systems.

Server-Class Expansion in an Embeddable Form Factor

LattePanda Sigma 2 provides extensive connectivity for storage, networking, and acceleration. It includes three M.2 M-Key slots, one M.2 E-Key slot, PCIe 4.0 x2 via OCuLink, and two Thunderbolt 4 ports. Dual 10GbE Ethernet supports high-bandwidth networking, while USB 3.2 Gen 2, USB 2.0, HDMI 2.1, DisplayPort 2.1, and eDP 1.5 provide flexible system integration options. These interfaces allow LattePanda Sigma 2 to operate as a compact local AI server or integrate directly into larger intelligent systems.

Ready for Local AI Development and Deployment

LattePanda Sigma 2 supports Windows 11 and Ubuntu 24.04 LTS or later, with compatibility for OpenVINO™, llama.cpp, and Ollama. It ships with Ubuntu, GPU and NPU drivers, OpenVINO Model Server (OVMS), OpenClaw, and local AI models pre-installed on the onboard SSD, reducing setup time for local AI deployment.

Designed for Product Integration

LattePanda Sigma 2 is designed for product-level integration and long-term deployment, offering a guaranteed 3+ year product lifecycle, custom boot logos and BIOS customization options, and direct engineering support for integration projects. Its combination of local AI compute, high-speed I/O, flexible expansion, and real-time control supports a wide range of applications, including:
Local AI Appliances — Run private RAG and AI agents while keeping sensitive data on premises.
All-in-One Homelabs — Combine NAS, virtualization, media services, and local AI in one system.
Professional Instruments — Capture and process high-speed data streams directly inside compact professional equipment.
Mobile Autonomous Robots — Unify perception, navigation, planning, and real-time control on one platform.

"Local AI is moving beyond development and into the products and systems people use every day," said Youliang Yu, Product Manager of LattePanda Sigma 2. "With LattePanda Sigma 2, we are combining the computing performance, memory capacity, expansion, and real-time hardware control needed to run AI locally in a compact, embeddable platform. The goal is to give developers a practical way to put local AI directly into the products they build."

Key Specifications

Processor: Intel^® Core™ Ultra X7 358H
AI Performance: Up to 180 TOPS (CPU + GPU + NPU, INT8)
NPU: Up to 50 TOPS
CPU: 16 cores (4P + 8E + 4LPE), up to 4.8 GHz, 18MB cache
GPU: Intel Arc™ B390, 12 Xe-cores, up to 122 TOPS (INT8)
Memory: Up to 64GB LPDDR5X-8533 RAM, 136.5 GB/s bandwidth
Storage & Expansion: 3× M.2 M-Key, 1× M.2 E-Key, OCuLink PCIe 4.0 x2
Networking: 2× 10GbE RJ45
Thunderbolt: 2× Thunderbolt 4 Type-C
USB Type-A: 2× USB 3.2 Gen 2, 2× USB 2.0
Display: HDMI 2.1, DP 2.1 via USB Type-C, eDP 1.5
Co-Processor: RP2350
Operating System: Windows 11, Ubuntu 24.04 LTS or later
Dimensions: 146 × 102 mm

For more information about LattePanda Sigma 2, please visit the LattePanda website. During the pre-sale period through mid-November, the LattePanda Sigma 2 is available at a promotional price of $2,599. Following the conclusion of the pre-sale period, the price will return to $2,799.

View original content to download multimedia:https://www.prnewswire.com/de/pressemitteilungen/lattepanda-launches-lattepanda-sigma-2-the-embeddable-ai-single-board-server-302916499.html