← All hardware
In-memory compute edge accelerator: 214 TOPS at 14.7 TOPS/W
Pros
- Excellent efficiency
- Strong TOPS for edge
- Flexible form factors
- Good for multi-stream vision
Cons
- Inference-only
- INT8-centric (limited GenAI)
- Smaller ecosystem
- Vision-focused
✓ Where it shines / best for
- Edge AI computer-vision deployments (smart retail, industrial, cities)
- Multi-camera real-time video analytics on servers/edge boxes
- Developers needing high TOPS-per-dollar inference acceleration
✕ Not the best fit for
- LLM/generative training workloads
- Cloud-scale data-center training (it is an inference accelerator)
- Consumers wanting plug-and-play (requires integration/SDK work)
Features
- ✓ On-device / offline
- ✓ Edge AI
- ✓ Real-time
- ✓ Computer Vision
- ✓ Inference Accelerator
- ✓ PCIe
- ✓ In Memory Compute
- ✓ SDK
- ✓ Video Analytics
Pricing
| Plan | Price | Billing | Notes |
|---|---|---|---|
| Metis PCIe AI Acceleration Card | ~$199 | one-time | Single Metis AIPU PCIe card; pricing via Axelera/distributors, evaluation systems higher |
| Metis Compute Board / dev kit | Contact vendor | one-time | Bundled developer/evaluation systems priced separately |
Pricing verified from the official source. Prices change often — confirm on the vendor's site before buying.
Specifications
| use | Edge vision inference |
| power | ~Few watts to ~tens of watts |
| memory | 1GB+ DRAM on module |
| performance | 214 TOPS INT8, 14.7 TOPS/W |
| architecture | Digital in-memory compute (quad-core) |
Sponsored
A full review is being generated for this product and will appear here shortly.