Google Coral USB Edge TPU Accelerator — Edge AI
Google Coral USB Accelerator with Edge TPU coprocessor. 4 TOPS INT8, USB 3.0 Type-C, fanless. For Raspberry Pi, Linux, Windows. On-device ML inference at the...
Edge AI
Inference
Hailo-8 M.2 AI Accelerator — 26 TOPS Edge AI Module
Hailo-8 M.2 AI Acceleration Module. 26 TOPS at sub-15W, M.2 2280 M-Key form factor, PCIe Gen3 x4. For edge AI inference on Raspberry Pi 5, Jetson, industrial...
Edge AI
Inference
NVIDIA Jetson AGX Orin 64GB — Edge AI Developer Kit
NVIDIA Jetson AGX Orin Developer Kit 64GB. 275 TOPS, 2048-core Ampere GPU, 12-core ARM Cortex-A78AE, 32GB LPDDR5. For autonomous machines, robotics, edge AI.
Edge AI
Inference
NVIDIA Jetson AGX Thor — Blackwell Edge AI Developer Kit
NVIDIA Jetson AGX Thor Developer Kit with T5000 module. 2,070 FP4 TFLOPS, 2560-core Blackwell GPU, 14-core ARM Neoverse V3AE, 128GB LPDDR5X. For humanoid rob...
Edge AI
Inference
NVIDIA Jetson Orin Nano 8GB Dev Kit — 40 TOPS · 6-core ARM · 8GB LPDDR5
NVIDIA Jetson Orin Nano 8GB Developer Kit — the most affordable entry point to NVIDIA edge AI. 40 TOPS, 8GB LPDDR5, Ampere GPU with 1,024 CUDA cores. Include...
Edge AI
Inference
NVIDIA Jetson Orin Nano Super 8GB — Edge AI Dev Kit
NVIDIA Jetson Orin Nano Super Developer Kit 8GB. 67 TOPS, 1024-core Ampere GPU, 6-core ARM A78AE. Compact, low-cost edge AI for hobbyists and prototyping.
Edge AI
Inference
NVIDIA Jetson Orin NX 16GB — Edge AI Module
NVIDIA Jetson Orin NX 16GB AI module. 100 TOPS, 1024-core Ampere GPU, 8-core ARM A78AE, 16GB LPDDR5. 10W-25W configurable. Compact edge AI for drones, camera...
Edge AI
Inference
NVIDIA Jetson Orin NX 8GB — 70 TOPS · 6-core ARM · 8GB LPDDR5
NVIDIA Jetson Orin NX 8GB delivers 70 TOPS AI performance in a compact SO-DIMM form factor. 6-core ARM Cortex-A78AE CPU, 8GB LPDDR5, Ampere GPU with 1,024 CU...
Edge AI
Inference
MemryX MX3 — M.2 AI Accelerator Module · 96 TFLOPS Pipelined
MemryX MX3 — PCIe Gen3 M.2 2280 M-key AI accelerator module. Up to 4 chips, 6 TFLOPS @1GHz per chip (96 TFLOPS pipelined), 0.6-2W per chip, -40 to +85°C. ONN...
Edge AI
Inference
Rockchip RK3588S AI Dev Board — 6 TOPS NPU · Octa-core ARM · 16GB · 8K Decode
Rockchip RK3588S AI development board with 6 TOPS NPU, octa-core ARM CPU (4×A76 + 4×A55), 16GB LPDDR4X, 8K video decode, and rich I/O. Cost-effective alterna...
Edge AI
Inference
Virtium Edge AI — VMX-004 M.2 Accelerator & RZ/G3E DX-M1 SOM
TAA-compliant, US-assembled M.2 2280 AI accelerator module with 4× MemryX MX3 for up to 24 TOPS at 3–10 W, and the RZ/G3E DX-M1 SOM at 25 TOPS...
Edge AI
Inference
Black Sesame Huashan A2000 — Jiushao-NPU Automotive AI SoC
High-performance L2–L5 automated driving AI chip platform with a large-core Jiushao NPU, FP16/FP8/INT4/INT8/INT16 precision and the BaRT toolchain. Cleared by US DoC/DoD in January 2026 for global sale.
Edge AI
Automotive
Qualcomm Dragonwing IQ-2390 — Industrial Edge AI Processor
First Dragonwing IQ2 Series processor. Quad-core Kryo, Adreno 704 GPU, RISC-V real-time MCU, dual Gigabit Ethernet with TSN, and a -30C to +115C extended industrial range.
Industrial
TSN
Edge AI
ESWIN EIC7700X — RISC-V Edge AI SoC
4x SiFive P550 RISC-V cores at 1.8GHz with a 19.95 TOPS INT8 NPU, Imagination AXM-8-256 GPU, 32GB LPDDR5, 8K H.265 codecs and roughly 8W typical power.
RISC-V
19.95 TOPS
Edge AI
Horizon Robotics Journey 6 — Automotive AI SoC
Automotive edge AI family scaling from 10 to 560 TOPS. Flagship Journey 6P carries a 4-core Nash BPU, an 18-core CPU at 410K DMIPS and 37B transistors on TSMC 7nm.
560 TOPS
Automotive
ADAS
Waymo carTPU — 5nm Sensor Fusion Processor
160 TOPS INT8 and 80 TFLOPS FP16 in a 45 mm FCBGA under 75 W, with in-package LPDDR5x at 273 GB/s.
Automotive
Sensor Fusion
BOS Semiconductors Eagle-N — 250 TOPS Automotive AI Chiplet SoC
250 TOPS INT8 via Tenstorrent Tensix chiplets, scaling past 2,000 TOPS. ASIL-B and AEC-Q100 Grade 2.
Automotive
Chiplet
Synaptics Astra SL2610 — Multimodal GenAI Edge Processor Line
Five AI-native SoC families on Cortex-A55 and Cortex-M52 Helium cores with transformer-capable NPUs and the Torq Edge AI platform running Google’s Coral RISC-V NPU.
GenAI Edge
Coral NPU
Ambarella X7 — Standalone CVflow Edge AI Accelerator
Ambarella’s first dedicated CVflow accelerator chip adds vision-transformer inference to any Arm or x86 host over one PCIe Gen 3 lane or USB 3.2.
PCIe Gen 3
2–5 W
Qualcomm Dragonwing IQ10 — 350 TOPS Industrial Edge AI
18-core Oryon CPU with NPU/GPU, up to 350 TOPS dense, Safety Island ECC, 12x GMSL2 cameras, TSN/EtherCAT/8x CAN-FD for AMRs and humanoid robotics.
Edge AI
Robotics
Qualcomm Dragonwing IQ-X Series — 45 TOPS Industrial SoC
12-core Oryon at 3.4 GHz, Hexagon NPU up to 45 TOPS, Adreno GPU, up to 64 GB LPDDR5x, -40 C to 105 C in a 58 mm FCBGA for PLCs, HMIs and box PCs.
45 TOPS
Industrial SoC
Quadric Chimera GPNPU — Programmable Neural Processor IP
Licensable GPNPU IP blending NPU throughput with full C++ programmability: 512 MAC to 8K MAC cores, 1 to 6,912 TOPS, INT8 plus FP16, 1-16 MB configurable L2 and one ONNX-plus-C++ toolchain.
1-6,912 TOPS
C++ Programmable
DEEPX DX-M2 — 80 TOPS Generative AI NPU
80 TOPS INT8 inside a 5 W budget, running 20–100 B parameter mixture-of-experts LLMs at 20–30 tokens/s on-device.
80 TOPS
5 W
MoE LLM
AICore DX-M1M — 25 TOPS M.2 2242 Module
DEEPX DX-M1M SoC on a short M.2 2242 M+B Key card: 25 TOPS, 1 GB LPDDR4X, PCIe Gen3 ×2 at ~3 W typical.
25 TOPS
M.2 2242
3 W
DEEPX DX-M1+ — 50 / 100 TOPS Multi-Chip Module
Two or four DX-M1 NPU chips on one M.2 board — 50 or 100 TOPS, with a dual-chip variant adding Rockchip video transcode.
50–100 TOPS
Multi-chip
Transcode
Quectel SP895BD-AP — AIoT Module on Qualcomm Dragonwing Q-8750 with 80 TOPS NPU
Quectel SP895BD-AP smart AIoT module on Qualcomm Dragonwing Q-8750. 8-core Oryon CPU to 4.32GHz, 80 TOPS NPU, Adreno Series 8 GPU, triple ISP for 3x48MP cameras, 8K encode/decode, Android 15 or Linux, compact LGA.
80 TOPS
AIoT Module
8K Video
Zhihe A210 — 12 TOPS RISC-V Edge AI SoC
Octa-core RISC-V RV64GCV processor with a 12 TOPS INT8 NPU, INT4 to BF16 datatype coverage, 4Kp120 AV1 decode and 8 tokens/s DeepSeek-7B edge inference.
RISC-V
12 TOPS
Edge AI
Allwinner T536 — Industrial Edge AI SoC with ECC RAM
Quad Cortex-A55 plus RISC-V cores with inline ECC DRAM support, a 3 TOPS NPU, dual Gigabit Ethernet, 4x CAN-FD and a -40 to +85C industrial temperature range.
Industrial SoC
ECC RAM
3 TOPS
Amlogic A123X and C305X2 — Cortex-A320 AIoT SoCs
Armv9 Cortex-A320 AIoT processors pairing a 4 TOPS ADLA2 vision NPU with an 8 TOPS ADLA3 transformer NPU, AI-ISP to 4Kp30 and 4Kp60 encode on TSMC 6nm.
Cortex-A320
8 TOPS NPU
AI-ISP
Quectel QSM603FCP — MediaTek MT8371 Edge AI Smart PCBA
MediaTek MT8371 octa-core Cortex-A78/A55 smart PCBA with Mali-G57 MC2 GPU, Cat 4 LTE, Wi-Fi 6, Bluetooth 5.4 and MIPI CSI, HDMI, RS485 and CAN I/O.
MT8371
Cortex-A78/A55
Cat 4 LTE
Quectel SE200ZC-AP — Rockchip RV1126B Vision Smart Module
Rockchip RV1126B smart module with quad-core Cortex-A53, 12M ISP HDR, 8M AI-ISP and a 3 TOPS INT4/8/16 and FP16 NPU with 4K H.265 encode.
RV1126B
3 TOPS NPU
4K H.265
Mobilint REGULUS MLD-R1 AI USB — 10 TOPS
REGULUS NPU USB 3.1 Gen1 stick, 10 TOPS INT8 at 2.4 W, 4/8 GB LPDDR4X, IP65. x86/ARM/RISC-V hosts.
10 TOPS
USB
2.4W
Innatera Pulsar — Neuromorphic SNN Processor
Spiking neural network fabric + CNN accelerator + RISC-V core in 2.8x2.6 mm at milliwatt power. Always-on sensing.
Neuromorphic
SNN
mW
TechNexion TELOS-AI4000 — NXP Ara240 40 TOPS Module
Enclosed edge-AI accelerator module with 40 eTOPS, 16 GB LPDDR4, PCIe host interface and 12 W typical power.
40 TOPS
Ara240
12 W
Forlinx RK1820 / RK1828 M.2 — 20 TOPS
M.2 2280 AI accelerator card with 20 TOPS INT8, 2.5 GB / 5 GB 3D-stacked DRAM and 100+ tokens/s offline LLM inference.
20 TOPS
M.2 2280
LLM
GigaDevice GD32H7 Series — 600 MHz Cortex-M7 Edge AI MCUs
600 MHz Arm Cortex-M7 MCU family (GD32H75E/H759/H757/H737) with up to 4 MB flash, 1 MB SRAM and an EtherCAT SubDevice controller for on-device edge AI and real-time control.
Edge AI
Cortex-M7
Huawei Atlas 200 — Ascend 310 AI Accelerator Module
Compact Ascend 310 accelerator module delivering 16 TOPS INT8 and 8 TOPS-class FP16 with 8GB LPDDR4X at ~9.5W, for cameras, drones, robots and edge video devices.
Edge AI
Ascend 310
Texas Instruments Edge AI Processors — Jacinto and AM6xA SoCs
TDA4VM (8 TOPS MMA, ASIL-D), AM68A and AM62A vision processors with the TIDL software stack scale from one to eight cameras for machine vision and robotics.
Edge AI
TDA4VM
Sony IMX500 / IMX501 AITRIOS — Intelligent Vision Sensors with On-Chip AI
Stacked 12.3 MP intelligent vision sensor with on-chip AI DSP and model memory: 30 fps inference and metadata-only output, running the AITRIOS platform without an external processor.
Edge AI
IMX500
Hailo-8R mPCIe — Rugged 13 TOPS AI Module
Rugged Mini PCIe AI accelerator with 13 TOPS INT8 in a Full-Mini F1 card, -40 to +85C range and PCIe Gen 3.0 x1 host interface.
Edge AI
mPCIe
13 TOPS

All 69 Models

Complete list of Edge AI Accelerators & Modules — click any model for full specifications, pricing, and official manufacturer part numbers (MPN).

Google Coral USB Edge TPU Accelerator — Edge AI · Edge AI Hailo-8 M.2 AI Accelerator — 26 TOPS Edge AI Module · Edge AI NVIDIA Jetson AGX Orin 64GB — Edge AI Developer Kit · Edge AI NVIDIA Jetson AGX Thor — Blackwell Edge AI Developer Kit · Edge AI NVIDIA Jetson Orin Nano 8GB Dev Kit — 40 TOPS · 6-core ARM · 8GB LPDDR5 · Edge AI NVIDIA Jetson Orin Nano Super 8GB — Edge AI Dev Kit · Edge AI NVIDIA Jetson Orin NX 16GB — Edge AI Module · Edge AI NVIDIA Jetson Orin NX 8GB — 70 TOPS · 6-core ARM · 8GB LPDDR5 · Edge AI MemryX MX3 — M.2 AI Accelerator Module · 96 TFLOPS Pipelined Rockchip RK3588S AI Dev Board — 6 TOPS NPU · Octa-core ARM · 16GB · 8K Decode · Edge AI Forlinx FCU3101 — 2 TOPS NPU InHand Mo 68A — 8 TOPS InHand Mo 62A — 2 TOPS Advantech AOM-6741 — Qualcomm IQ-9075 SMARC AI Vision Module, 100 TOPS Virtium VMX-004 — 24 TOPS M.2 2280 AI Accelerator, TAA / Made in USA CamThink NeoEyes NE302 — STM32N6 · Neural-ART NPU · WiFi 6 Ambarella N1-655 — Edge GenAI SoC for Multi-Channel VLM Under 20W EdgeCortix SAKURA-II — 60 TOPS 8W M.2 & PCIe Edge AI Accelerator NXP Ara240 (Kinara Ara-2) — 40 eTOPS M.2 Edge AI Module Geniatech AIM AI Accelerator Modules — Hailo-8/10 · NXP Ara-240 · MemryX MX3 · DEEPX Radxa AICore SG2300x — SOPHON SG2300x · 24 TOPS INT8 · 48 TOPS Cascade Lanner Falcon-H8 — Multi-Hailo-8 PCIe AI Acceleration Card · up to 156 TOPS Black Sesame Huashan A2000 — Jiushao-NPU Automotive AI SoC · L2-L5 Qualcomm Dragonwing IQ-2390 — Industrial Edge AI Processor · TSN · -30 to +115C ESWIN EIC7700X — Quad-Core RISC-V Edge AI SoC · 19.95 TOPS NPU Horizon Robotics Journey 6 — Automotive Edge AI SoC Family · 10-560 TOPS SiMa.ai Modalix MLSoC — 50 TOPS Sub-10W Edge Physical AI Platform Waymo carTPU — 5nm Sensor Fusion Processor BOS Semiconductors Eagle-N — 250 TOPS Automotive AI Chiplet SoC Synaptics Astra SL2610 — Multimodal GenAI Edge Processor Line Ambarella X7 — Standalone CVflow Edge AI Accelerator Qualcomm Dragonwing IQ10 — 350 TOPS Industrial Edge AI Robotics Platform Qualcomm Dragonwing IQ-X Series — 45 TOPS Industrial SoC (IQ-X7181, IQ-X5121) Axera AX8850 — 24 TOPS Edge AI Platform (AI Pyramid, LLM-8850, AICore AX-M1) Quadric Chimera GPNPU — 1 to 6,912 TOPS Programmable Neural Processor IP NXP Kinara Ara-2 Discrete NPU — GenAI Inference at the Edge Mythic M1076 Analog Matrix Processor — 25 TOPS in a Single Chip DEEPX DX-M2 — 80 TOPS Generative AI NPU at 5 W AICore DX-M1M — 25 TOPS M.2 2242 Edge AI Module on DEEPX DX-M1M DEEPX DX-M1+ — Multi-Chip M.2 AI Module, 50 TOPS or 100 TOPS Synaptics SYN765x — Wi-Fi 7 AI-Native Connected MCU with Ethos-U55 NPU Synaptics Astra SRW1500 Series — AI MCUs with Tri-Band Wi-Fi 7 and Ethos-U55 NPU Semidynamics Cervell — All-in-One RISC-V NPU IP from 8 to 256 TOPS Quectel SP895BD-AP — AIoT Module on Qualcomm Dragonwing Q-8750 with 80 TOPS NPU DEEPX DX-AIPlayer N97 — Intel N97 Mini PC with 25 TOPS DX-M1 NPU Upbeat UP201 / UP301 — Always-On RISC-V AI MCUs with Dual NPU and 2.5D GPU Zhihe A210 — 12 TOPS RISC-V Edge AI SoC Allwinner T536 — Industrial Edge AI SoC with ECC RAM Amlogic A123X / C305X2 — Cortex-A320 AIoT SoCs Quectel QSM603FCP — MediaTek MT8371 Edge AI Smart PCBA Quectel SE200ZC-AP — Rockchip RV1126B 3 TOPS Vision Smart Module Nordic nRF54LC10A — Bluetooth LE 5.4 + 802.15.4 SoC with Sub-50 nA Hibernation Realtek RTL8773J — Bluetooth 7-Ready Cortex-M55/M33 Audio MCU Virtium VMX-004 — Made-in-USA M.2 AI Accelerator Module on MemryX MX3 Texas Instruments Edge AI Processors — Jacinto and AM6xA Vision SoCs Sony IMX500 / IMX501 AITRIOS — Intelligent Vision Sensors with On-Chip AI · 12.3 MP · 30 fps inference · MIPI/SPI Flex Logix InferX X1 — 8.4 TOPS Edge AI Inference Accelerator Advantech MIC-741-AT — Jetson Thor Edge AI Inference System Qualcomm Dragonwing Q-2390 — Industrial AI Processor
Mobilint REGULUS MLD-R1 AI USB — 10 TOPS Innatera Pulsar Neuromorphic MCU TechNexion TELOS-AI4000 — 40 TOPS NXP Ara240 Edge AI Module Forlinx RK1820 / RK1828 M.2 — 20 TOPS Edge AI Accelerator GigaDevice GD32H7 Series — 600 MHz Cortex-M7 Edge AI MCUs Huawei Atlas 200 — Ascend 310 AI Accelerator Module Innodisk APEX-A100 — Dragonwing Edge AI System Qualcomm Dragonwing Q-7790 and Q-8750 IoT Processors Albatron.ai TALO-A1000 — Robotics & Autonomous AI Platform, up to 2070 TOPS Hailo-8R mPCIe — Rugged 13 TOPS Mini PCIe AI Acceleration Module