LattePanda Sigma 2 Image © LattePandaLattePanda Sigma 2 (Image © LattePanda)

The LattePanda Sigma 2 functions as an embedded AI server and combines three different computing elements—CPU, GPU, and NPU—to achieve its rated performance of 180 TOPS.

The technical breakdown of the AI performance is as follows:

  • GPU: The Intel Arc B390 features 12 Xe cores, which contribute up to 122 TOPS (INT8).
  • NPU: The integrated Neural Processing Unit delivers an additional 50 TOPS.
  • CPU: A 16-core processor (consisting of 4 performance cores, 8 efficiency cores, and 4 low-power efficiency cores) with clock speeds of up to 4.8 GHz.

LattePanda Sigma 2 CPULattePanda Sigma 2 CPU (Image © LattePanda)

To handle demanding local AI workloads, the system supports up to 64 GB of LPDDR5X-8533 RAM with a memory bandwidth of 136.5 GB/s. A key feature for AI developers is the ability to allocate up to 59 GB of VRAM. This capacity is sufficient to run quantized “Mixture-of-Experts” (MoE) models in the 35B class and to manage the KV cache required by AI agents with long context windows. Using INT4 quantization and OpenVINO, the system can run the “Qwen3.6-35B-A3B” model locally at approximately 30 tokens per second.

LattePanda Sigma 2 RAMLattePanda Sigma 2 RAM (Image © LattePanda)

The board was designed for server-class connectivity in a form factor suitable for embedded applications, allowing it to be integrated directly into a product rather than functioning as a standalone external unit.

LattePanda Sigma 2 PerformanceLattePanda Sigma 2 Performance (Image © LattePanda)

The system is equipped with two 10-GbE Ethernet ports to support high-bandwidth network environments. For external expansion and connecting peripherals, it features:

  • Thunderbolt 4: Two Type-C ports.
  • OCuLink: PCIe 4.0 x2 interface for adding AI accelerators.
  • M.2 slots: Three M.2 M-key slots and one M.2 E-key slot for storage and wireless expansions.
  • USB ports: Two USB 3.2 Gen 2 and two USB 2.0 ports.

For system integration and monitoring, the board offers HDMI 2.1, DisplayPort 2.1 via USB Type-C, and eDP 1.5.

Unlike traditional mini-PCs, the LattePanda Sigma 2 features an integrated RP2350 microcontroller. This coprocessor provides GPIO connectivity and enables the AI system to interact directly with physical sensors and actuators in real time. The board ships with pre-installed MicroPython firmware and offers a programmable interface for automation and the development of smart devices.

The platform supports Windows 11 and Ubuntu 24.04 LTS (or higher). To reduce deployment time, the system ships with Ubuntu, GPU and NPU drivers, the OpenVINO Model Server (OVMS), and OpenClaw. Several local AI models are preinstalled on the integrated SSD. The system is fully compatible with industry-standard tools such as llama.cpp, Ollama, and OpenVINO.

LattePanda Sigma 2 LLMsLattePanda Sigma 2 LLMs (Image © LattePanda)

Industrial Application and Lifecycle

The LattePanda Sigma 2 is designed for long-term professional use and offers a guaranteed product lifecycle of at least three years. For system integrators, the company offers options for BIOS customization as well as the ability to implement custom boot logos.

The hardware is geared toward various computationally intensive use cases:

  • Local AI appliances: Running private RAG (Retrieval-Augmented Generation) and AI agents to keep sensitive data on-site.
  • Autonomous robotics: Integration of perception, navigation, and real-time control on a single platform.
  • Professional measurement technology: Processing of high-speed data streams in compact measuring devices.
  • Home Labs: Combining NAS, virtualization, and local AI services.

Technical 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, 18 MB cache
  • GPU: Intel Arc™ B390, 12 Xe cores, up to 122 TOPS (INT8)
  • Memory: Up to 64 GB 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
  • Network: 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
  • Coprocessor: RP2350
  • Operating System: Windows 11, Ubuntu 24.04 LTS or higher
  • Dimensions: 146 × 102 mm

Pricing and Availability

The LattePanda Sigma 2 is currently available for pre-order.

Until mid-November, it will be offered at a promotional price of $2,599. After the pre-sale period ends, the regular retail price of $2,799 will apply again.