PCIe AI accelerator card. Powered by quad-core Metis AIPU | Axelera AI Store
Axelera AI
Overview
Axelera Metis PCIe AI accelerator card houses four Metis AIPU chips delivering up to 856 TOPS combined inference throughput via a standard PCIe x8 interface. At $379.95 with sub-75W TDP, it offers exceptional TOPS-per-dollar for edge server and workstation inference deployments. Supports ONNX model deployment with a Python SDK.
Key Features
- 4x Metis AIPU chips — up to 856 TOPS total
- Standard PCIe form factor (x8 interface)
- Sub-75W TDP — no auxiliary power required
- ONNX model deployment + Python SDK
- Dataflow AI architecture for high parallel throughput
Ideal For
AI engineers and enterprises deploying cost-effective inference acceleration for computer vision, NLP, and transformer models in standard x86 servers or workstations without GPU infrastructure.
No specifications available.
$379.95
Prices may vary. Verify on vendor site.
Tags
computer-visionedge-inferencellm-inference