NVIDIA has announced Groq 3 LPX, a new model designed to deliver ultrafast interactivity while handling extended context lengths on the NVIDIA Vera Rubin platform. The development aims to improve inference speeds for long-context applications, addressing performance demands in enterprise and research deployments.