[KongchangAI]
Company

Subquadratic

A research organization that developed the SubQ Snap decision model and the Subquadratic Sparse Attention (SSA) sparse attention architecture.

Timeline (last 90 days)

Oct 13

SubQ Snap is a decision model released by the research organization Subquadratic; it does not generate text, but instead returns a probability distribution for a given set of candidate answers.

Unverified50%
Oct 13

Six hosted decision models were released within three weeks: Jev, Perplexity's Decider, Cloudflare's Clef, OpenAI's Decisions API, Microsoft's Decision-1, and SubQ Snap.

Unverified50%
Oct 13

Subquadratic claims that SubQ Snap's Subquadratic Sparse Attention (SSA) can reduce attention cost from quadratic to linear, returning 100K tokens within 1 second, but these figures have no hardware specification or third-party verification.

Unverified50%
Oct 13

SubQ Snap officially claims to lead on 13 of 16 enterprise tasks, with phishing detection improving from 25.1 to 47.1 and code output prediction improving from 57.1 to 73.8.

Unverified50%
Oct 13

All of SubQ Snap's performance data comes from Subquadratic itself and has not been independently tested by an external lab, and its long-context tests are its own modified versions of RULER and LCR.

Unverified50%

All Facts (5)

Source Articles