TAAFT
Free mode
100% free
Freemium
Free Trial
Deals

SubQ 1M Preview

SubQ 1M-Preview is Subquadratic’s debut long-context model, built on a fully subquadratic architecture that scales linearly with context length. The company says this enables stronger long-context accuracy together with faster inference and lower cost, rather than forcing the usual tradeoff between scale and quality. It is being launched through an API, SubQ Code for full-repository coding, and SubQ Search for deep-research-style search. Subquadratic reports 95% on RULER 128K, a third-party-verified 65.9 on MRCR v2, 81.8 on SWE-Bench Verified, and research results up to 12 million tokens.
New Multimodal Gen 3
Released: May 5, 2026

Overview

SubQ 1M-Preview is Subquadratic’s first large language model and the first fully subquadratic LLM, built so compute grows linearly with context length rather than quadratically. It is designed for massive-context reasoning, coding, search, and full-corpus processing, making million-token windows more practical, faster, and cheaper than standard transformer-based approaches

About Subquadratic

Industry: Artificial Intelligence
Company Size: 16
Location: Wilmington, Delaware, US
Website: subq.ai
View Company Profile

Tools using SubQ 1M Preview

No tools found for this model yet.

Last updated: May 6, 2026
0 AIs selected
Clear selection
#
Name
Task