Is SubQ Snap 1.1 state of the art?Pending benchmarks.Announced Oct 22, 2026; no independent results yet.
Subquadratic's enterprise decision model: it reads supplied context, up to 384K tokens in a single pass, and returns structured answers such as a choice, score or label for routing, tool selection and classification. Built on the lab's proprietary Subquadratic Sparse Attention. Available through the API at subq.ai; weights not stated.
The details so far
- Lab
- Subquadratic
- Announced
- Oct 22, 2026
- API
- Live
- Open weights
- Unknown
- Type
- Multimodal decision model (subq.ai)
- Context
- 384K tokens in a single pass
- Architecture
- Subquadratic Sparse Attention (SSA), proprietary
- API price
- $0.04 per 1M input tokens, no charge for output; 100M free tokens for new users (CEO's launch post)
- Access
- subq.ai
- Weights
- Not stated
Tags marked unsure are not stated on the lab's own pages.
What would let us call it
Verdicts use only independent boards we already track. Two consecutive snapshots are required before any crown.
- Publicly announced with a source
- Usable by the public (API or weights)
- Listed on an independent board we track
- Enough results for a rank in a category
- Holds across two snapshots (crowns only)
What the lab claims
Self-reported. Shown for context; not used in any verdict.
Lab claim: tied for first on the Artificial Analysis Decision Index (65, with d1-ultra and Elephant 9B, per subq.ai).
Lab claim: 96.2 on When2Call, up from 75.2 (CEO's launch post); subq.ai's chart lists 91.0 on When2Call against 75.9 for Jev 1.14.
Lab claim: 7.2 points ahead of Jev across 16 enterprise benchmarks, and 81% fewer confident errors on questions it can't answer.
Sources
LABSubQ Snap 1.1 (subq.ai)subq.ai · Oct 22, 2026LABLaunch post on X by Subquadratic CEO Alexander Whedonx.com/alex_whedon/status/2113356853308469444 · Oct 22, 2026Tracking continues until an independent board lists it.