Groq LPU — specs & cloud pricing
The Groq LPU is a wafer-scale / sram inference from Groq. Live specs from the Mintok silicon catalog.
Specifications
- Vendor
- Groq
- Class
- Wafer-scale / SRAM inference
- Architecture
- SRAM
- BF16 throughput
- 750 TFLOPS
- On-package memory
- 0.23 GB
- Memory bandwidth
- 80,000 GB/s
- TDP
- 300 W
FAQ
How much memory does the Groq LPU have?
The Groq LPU has 0.23 GB of on-package memory at 80,000 GB/s bandwidth.
What is the Groq LPU's BF16 compute throughput?
The Groq LPU delivers 750 BF16 TFLOPS at a 300 W TDP.
Is the Groq LPU a datacenter or edge accelerator?
The Groq LPU is a wafer-scale / sram inference (300 W TDP).
Cost this chip in a fleet
Size a deployment on the Groq LPU, or see the full priced rack BOM it sits in.
Specs and modeled rates maintained by Mintok's catalog watchers from public sources. Not investment advice. Methodology.