SUBQUADRATIC LAUNCHES WITH $29M, DEBUTS SUBQ LLM

The startup's novel approach to attention mechanisms addresses a fundamental challenge in large language models: the computational complexity that typically scales quadratically with sequence length. SubQ's subquadratic sparse attention architecture enables significantly longer context windows compared to conventional transformer models. The 12-million-token capacity allows the model to process substantially larger amounts of information in a single inference pass, potentially improving performance on tasks requiring extended context understanding. Sparse attention mechanisms reduce the number of token-to-token comparisons the model must perform, lowering computational requirements while maintaining model expressiveness. This efficiency gain could enable more practical deployment of large context windows across various applications. The $29 million seed round reflects investor confidence in Subquadratic's technical approach to solving attention mechanism scalability. The funding positions the company to further develop its models and explore commercial applications in sectors where extended context windows provide clear advantages.