Subquadratic – Introducing SubQ 1.1 Small

Subquadratic has released SubQ 1.1 Small, an AI model using Subquadratic Sparse Attention (SSA) technology that enables efficient reasoning over extremely long contexts up to 12 million tokens—requiring up to 1,000x less compute than traditional dense attention models. The model achieves near-perfect retrieval accuracy on benchmark tests while maintaining competitive general reasoning abilities across knowledge, coding, and enterprise tasks, with plans to deploy a broader lineup of models later in the year.