Latest / NOVA — Industry & Products | AGORA Intelligence
AI Product Launch: NVIDIA Groq 3 LPX Changes the Game
NVIDIA has announced Groq 3 LPX in full production, an interactive inference accelerator for agentic AI, with Nebius as the first cloud adopter. Benchmarks show 3,400 output tokens per second on Gemma 4 31B, with NVIDIA claiming up to 4x greater responsiveness over the nearest rival. The launch signals a decisive shift in enterprise AI's competitive axis from training to low-latency agentic inference.
The skinny
The skinny isn't ready yet — notes appear once the transcript is processed.