system-one-bench

~/system-one-bench / compare/patrick-18-dexy-decision-9b-vs-scaledown-classify-1

Head to head

Dexy-Decision-9B vs Classify-1

Dexy-Decision-9B scores 84.7 and Classify-1 30.6 on the System One Index. Dexy-Decision-9B is more accurate on 15 of 16 text datasets, Classify-1 on 1. Same cases, same request, same scorer.

Overall

Dexy-Decision-9BClassify-1
System One Index84.730.6
English82.330.7
Arabic86.830.6
Vision Index––
Calibration error3.5%3.2%
Median latency–254 ms
Cost per 1,000 calls–$0.018

Text datasets, accuracy

Dexy-Decision-9BClassify-1
AG News English91.526.5
SST-2 English95.550.0
SST-5 English60.013.0
DAIR Emotion English54.059.0
Banking77 English93.01.0
BoolQ English89.538.0
XNLI English English94.730.7
Typed decisions English80.027.2
XNLI French French89.330.7
XNLI Arabic Arabic89.330.7
MASSIVE scenario Arabic89.53.0
MASSIVE intent Arabic91.53.5
OSACT4 offensive Arabic91.580.0
Tweet sentiment Arabic67.035.0
MIRACL rerank Arabic88.012.0
MIRACL pairwise Arabic91.050.0

Model pages: Dexy-Decision-9B · Classify-1. Latency for API models is the median per call through OpenRouter, timed from a cloud server with 8 calls in flight.