Groq LLaMA

Groq LLaMA pairs Meta's open-weights LLMs with Groq's LPU Inference Engine to deliver industry-leading token throughput and ultra-low latency.

Groq LLaMA represents the integration of Meta's premier open-weights models (including Llama 3.1, Llama 3.3, and Llama 4) with Groq's custom Language Processing Unit (LPU) architecture. By bypassing traditional GPU bottlenecks, this setup achieves blazing-fast inference speeds (often exceeding 500 tokens per second for smaller models) and predictable, deterministic performance. Developers access these accelerated models via the GroqCloud Console, utilizing an OpenAI-compatible API to easily power real-time agentic workflows, complex tool-use tasks, and highly responsive conversational applications.

https://groq.com