Miami startup Subquadratic claims 1,000x AI efficiency gain with SubQ model; researchers demand independent proof

Miami startup Subquadratic says its SubQ 1M-Preview is the first large language model built on a fully subquadratic architecture.

The company claims the approach scales compute linearly with context and cuts attention compute by nearly 1,000x at 12 million tokens.

Subquadratic also launched private-beta products, SubQ Code and SubQ Search, and disclosed $29 million in seed funding.

AI researchers are skeptical and are urging independent verification before accepting the efficiency claims.