Groq LPU

by Groq · paid · Last verified 2026-04-24

Groq's Language Processing Unit (LPU) is a deterministic ASIC architecture optimized for sequential transformer inference, eliminating the memory-bandwidth bottlenecks of GPU-based serving. Groq LPU clusters deliver measured token generation speeds of 500+ tokens/second for Llama-class models, significantly outpacing GPU inference for latency-critical applications.