headquarters

San Francisco, US

SubQ is developing a sub-quadratic large language model architecture, enabling AI systems to reason across tens of millions of tokens of context at a fraction of the compute cost of traditional architectures.