Groq is an AI inference platform powered by its custom Language Processing Unit (LPU), delivering the fastest publicly available LLM inference. Run Llama 3, Mixtral, and Gemma at speeds up to 10× faster than GPU-based cloud providers — ideal for real-time AI applications.