Raspberry Pi AI HAT+ 2 (Hailo-10H 40 TOPS + 8GB RAM)
Second-generation official Raspberry Pi 5 GenAI HAT pairing the 40-TOPS (INT4) Hailo-10H neural processor with 8GB of dedicated onboard LPDDR4X memory—running 1.5B–3B local LLMs and VLMs at 3 Watts without touching host Pi RAM.
Key Engineering Highlights
- Includes 8GB of dedicated LPDDR4X RAM wired directly to the Hailo-10H NPU, leaving 100% of the Raspberry Pi 5's host memory and quad-core CPU free
- Generates ~9.5 to 11.8 tokens/sec on 3B INT4 language models and sub-400ms Time-to-First-Token at just 3 Watts of NPU power
- Drop-in `hailo-ollama` and OpenAI-compatible REST API server for offline Home Assistant voice pipelines and ROS 2 mobile robots
Full Hardware Specifications
| Processor / Edge SoC | Hailo-10H GenAI Dataflow NPU (40 TOPS INT4 / 20 TOPS INT8) + 8GB LPDDR4X |
|---|---|
| AI Model Stack | Llama-3.2-3B / Qwen2.5-Coder-3B / Whisper-Base / InternVL-2B |
| Display & Optics | Pass-Through GPIO Header + Raspberry Pi 5 Dual 4K60 HDMI |
| Camera & Vision | Direct Zero-Copy DMA from Raspberry Pi 5 Dual MIPI CSI-2 Ports |
| Measured Weight | 48.0g (HAT+ Module w/ Anodized Thermal Spreader) |
| Battery & Runtime | Powered via Raspberry Pi 5 PCIe FPC + GPIO (~3.0W Typical NPU Draw) |
| Connectivity & I/O | PCIe Gen 3 x1 Single-Lane FPC Ribbon (8 GT/s) |
| Sensors & Acoustics | Dedicated 8GB On-HAT LPDDR4X Model Weight & KV-Cache Buffer |
⚖️ Side-by-Side Comparison: Raspberry Pi AI HAT+ 2 (Hailo-10H 40 TOPS + 8GB RAM) vs. Edge Compute & DevKits Peers
Direct specification comparison against leading competing devices in the Edge Compute & DevKits category.
| Specification | Raspberry Pi AI HAT+ 2 (Hailo-10H 40 TOPS + 8GB RAM) (Current) | NVIDIA Jetson Orin NX Super Edge AI DevKit → | Xiaozhi AI Voice Cube Pro (ESP32-S3 Open DevKit) → | Apple Mac mini (M5 Pro 64GB) Private AI Node → |
|---|---|---|---|---|
| Brand | Raspberry Pi × Hailo | NVIDIA | Open Hardware Community | Apple |
| Retail Price | $130 | $499 | $12 | $1,799 |
| Lab Rating | ||||
| Edge SoC | Hailo-10H GenAI Dataflow NPU (40 TOPS INT4 / 20 TOPS INT8) + 8GB LPDDR4X | NVIDIA Ampere GPU (1024 CUDA + 32 Tensor Cores) | Espressif ESP32-S3R8 (Dual-Core LX7 @ 240MHz + 8MB PSRAM) | Apple M5 Pro (14c CPU + 20c GPU + 16c Neural Engine) |
| AI Model Stack | Llama-3.2-3B / Qwen2.5-Coder-3B / Whisper-Base / InternVL-2B | Runs Llama-3.2 8B / Qwen2.5-VL 7B / YOLO11 Locally | Edge ESP-SR Wake + Realtime Cloud Voice LLMs | MLX Accelerated Qwen3-32B / Llama-3.3-70B |
| Display / Optics | Pass-Through GPIO Header + Raspberry Pi 5 Dual 4K60 HDMI | HDMI 2.1 / DP 1.4 (Up to 8K) | 1.28-inch Circular IPS LCD (240×240) / 0.96" OLED | Supports up to three 6K displays |
| Camera | Direct Zero-Copy DMA from Raspberry Pi 5 Dual MIPI CSI-2 Ports | Dual MIPI CSI-2 Camera Lanes + 4× USB 3.2 | Optional OV2640 / GC2145 Camera Module | External USB-C / Studio Display Camera |
| Weight | 48.0g (HAT+ Module w/ Anodized Thermal Spreader) | 450g (including active heatsink & carrier board) | 62g (with 3D-printed enclosure & speaker cavity) | 730g |
| Battery | Powered via Raspberry Pi 5 PCIe FPC + GPIO (~3.0W Typical NPU Draw) | 19V DC Input (15W–40W configurable TDP) | 800mAh LiPo (5h continuous dialogue) | Internal 155W PSU (4W idle draw) |
📚 Related Engineering Teardowns & Reviews
Browse All 88 Articles →Raspberry Pi AI HAT+ 2 (Hailo-10H 8GB) Review & Benchmark: Running 3B Local LLMs and Whisper in 3 Watts for $130
RECIWO Labs tested the official $130 Raspberry Pi AI HAT+ 2 powered by the 40-TOPS Hailo-10H GenAI NPU and 8GB of dedicated on-HAT LPDDR4X memory, benchmarking Llama-3.2-3B, Qwen2.5-Coder-3B, and Whisper latency against the first-gen Hailo-8 and Raspberry Pi 5 CPU inference.
RayNeo X3 Pro vs Meta Ray-Ban Display Optical Teardown: Binocular 6,000-Nit MicroLED vs Monocular Geometric Waveguide
RECIWO Labs disassembled the 76g RayNeo X3 Pro binocular full-color MicroLED waveguide glasses ($1,099) and the 39g Baidu Xiaodu AI Glasses Pro ($299), benchmarking eye-box luminance, 6DoF spatial SLAM drift, and thermal dissipation against the $799 Meta Ray-Ban Display.
Tenstorrent Blackhole p150a Benchmark: Can a $1,399 Open RISC-V + Tensix Card With 210MB SRAM Challenge CUDA GPUs for Local LLMs?
RECIWO Labs tested Tenstorrent's $1,399 Blackhole p150a PCIe workstation accelerator (140 Tensix cores, 16 SiFive X280 Big RISC-V cores, 32GB GDDR6, 210MB on-chip SRAM, and 4x 800GbE QSFP-DD fabric ports), benchmarking open-source TT-Metalium and vLLM performance from single-card 32B models to switchless 4-card 128GB meshes.
AgiBot Lingxi X2 vs EngineAI PM01 vs Unitree R1 Benchmark: ViLLA Foundation Models, 320-Degree Waist Kinematics & Straight-Knee Gait RL
RECIWO Labs evaluated the three defining mid-size bipedal humanoids of late 2026: AgiBot's 35kg Lingxi X2 powered by the Genie Operator-1 (GO-1) ViLLA foundation model, EngineAI's 40kg open-source PM01 with its 320-degree rotating waist and 2.0 m/s straight-knee human gait, and Unitree's $5,900 25kg R1.
