Hailo-8 AI Processor
Delivering data center class performance to edge devices
Important Information:
Delivering data center class performance to edge devices

The Hailo-8 Family of Edge AI processors delivers data center-class AI performance in a compact, power-efficient form factor, enabling intelligent devices to perform real-time deep learning inference directly at the edge. With up to 26 TOPS of AI compute while consuming as little as 2.5 W, the Hailo-8 family is ideal for embedded vision, industrial automation, robotics, smart cities, intelligent transportation, retail analytics and security applications that demand high-performance AI with low latency. Built on Hailo's patented dataflow architecture, the Hailo-8 family is uniquely designed to maximize neural network efficiency without relying on external DRAM. By integrating all required memory directly on the processor, it simplifies hardware design, reduces system cost and supply chain risk, and delivers reliable, high-bandwidth AI processing. The scalable architecture also supports multi-stream and multi-model inference, as well as multiple Hailo-8 devices for higher-performance applications.
Available in a range of deployment options, including Hailo-8, Hailo-8L and Hailo-8R Series, the portfolio offers flexibility for rapid integration into commercial, industrial and automotive platforms. Supported by the comprehensive Hailo AI Software Suite with TensorFlow, PyTorch, ONNX and other leading AI frameworks, developers can efficiently optimize, deploy and scale AI applications from prototype to production.
Product Selection Guide ► About Hailo►
Buy Now

Hailo offers a holistic approach to AI at the edge, with a range of AI accelerators and AI camera SoCs with a single robust software suite, enabling flexibility and modularity in AI deployments.
AI Accelerators are Co-processors that integrate with edge platforms, empowering real-time deep learning inference tasks with low power consumption and high cost-efficiency. Supporting a wide range of neural networks, vision transformer models, and LLMs.
AI Vision Processors (VPUs) have AI-centric camera SoCs with high-performance AI processing supporting AI image enhancement, AI-driven video pipelines, and rich analytics in a camera power and cost envelope.
Feature |
Hailo-8 |
Hailo-8L |
Hailo-8R mPCIe Module |
Positioning | High-performance Edge AI Processor | Entry-Level Edge AI Processor | Ruggedized AI Acceleration Module |
AI Performance | 26 TOPS | 13 TOPS | 13 TOPS |
Typical Power Consumption | 2.5W | 1.5W | Best-in-class power efficiency |
External DRAM Required | No – Fully Integrated Memory | No – Fully Integrated Memory | Based on Hailo AI architecture, optimized for edge deployment 3 |
Multi-Stream Processing | Yes | Yes | Yes |
Multi-Model Inference | Yes | Yes | Yes |
AI Framework Support | TensorFlow, TensorFlow Lite, Keras, PyTorch, ONNX | TensorFlow, TensorFlow Lite, Keras, PyTorch, ONNX | TensorFlow, TensorFlow Lite, Keras, PyTorch, ONNX |
Host Platforms | x86, ARM | x86, ARM | x86, ARM |
Supported OS | Linux, Windows | Linux, Windows | Linux, Windows |
Operating Temperature | Industrial: -40°C to +85°C; Automotive: -40°C to +105°C | -40°C to +85°C | -40°C to +85°C |
Form Factor | Processor / M.2 / PCIe implementations available | Processor / M.2 / PCIe implementations available | Mini PCIe (mPCIe) Module |
Interface | PCIe-based solutions available | PCIe-based solutions available | PCIe Gen3 x1 (Mini PCIe Form Factor) |
Certification | Automotive-grade options available | Industrial-grade | CE, FCC |
Hailo delivers multiple innovations, addressing the fundamental properties of neural networks:

Deep Learning Development Frameworks
Published: 10th July 2026