#12m tokens

Miami Startup SubQ Claims 1000x Efficiency With Sub-Quadratic Model

Subquadratic emerged from stealth with a 12M-token context model it says breaks the computational scaling constraint that has limited every major LLM.