SubQ Snap 1.1 ties for first on the Artificial Analysis Decision Index

Source · Model card: SubQ Snap

Subquadratic released SubQ Snap 1.1 on 22 October 2026 (post; @alex_whedon, 16:48 BRT), ten days after the first SubQ Snap. It is still a closed, hosted model with 384K tokens of context, now at $0.04 per 1M input tokens (was $0.05), output free.

Independent result. Artificial Analysis scored Snap 1.1 at 65 on its Decision Index v1.0, tied with d1-ultra and Elephant 9B from the mammoth-1 family, 8 points ahead of Jev 1.14 (57). AA says the result comes mostly from a low hallucination rate: Snap 1.1 picked “I don’t know” on 86% of the knowledge questions it did not get right. Its accuracy on those questions is 20% (d1-ultra 32%, Elephant 9B 12%). AA measured $30 per 1M tasks, about 3.6× Elephant 9B for the same score.

Vendor numbers (UnverifiedClaim): +7.2 points over Jev 1.14 across 16 enterprise benchmarks, When2Call up from 75.2 to 96.2, calibration error down from 0.108 to about 0.014.

Snap 1.1 is not on the Hugging Face Decision Index or JevBench yet. The catalog score is unchanged.