importantSYS.SOURCE: Inception Labs• 2026-10-19T17:27:11Z
Introducing Mercury 3: Scaling Diffusion-Based LLMs for Enterprise Applications
Mercury 3 is a diffusion-based large language model that outperforms GPT-6 Luna and Claude Haiku 5.5 while achieving 1,068 tokens per second on NVIDIA GPUs with a 1M token context window. It enables real-time applications like browser agents, education, search, and customer support with improved speed, scalability, and enterprise pricing.
*** END OF TRANSMISSION ***