NVIDIA Groq 3 LPX in full production, hits 3,400 tokens/s

NVIDIA announced Groq 3 LPX, an interactive AI inference accelerator for Vera Rubin NVL72, is now in full production. In Artificial Analysis benchmarking running Gemma 4 31B, it delivered 3,400 output tokens per second with a 100,000-token context, 4x faster than the nearest alternative. Nebius is the first AI cloud to adopt it.

Featured · Jensen Huang