system-one-bench

~/system-one-bench / compare/patrick-18-dexy-decision-9b-vs-typesafe-jev-1-13

Head to head

Dexy-Decision-9B vs Jev 1.13

Dexy-Decision-9B scores 84.7 and Jev 1.13 79.0 on the System One Index. Dexy-Decision-9B is more accurate on 12 of 16 text datasets, Jev 1.13 on 3, and 1 are tied. Same cases, same request, same scorer.

Overall

Dexy-Decision-9BJev 1.13
System One Index84.779.0
English82.378.5
Arabic86.879.9
Vision Index–83.9
Calibration error3.5%6.1%
Median latency–150 ms
Cost per 1,000 calls–$0.028

Text datasets, accuracy

Dexy-Decision-9BJev 1.13
AG News English91.588.5
SST-2 English95.594.5
SST-5 English60.062.0
DAIR Emotion English54.056.0
Banking77 English93.079.0
BoolQ English89.589.5
XNLI English English94.786.0
Typed decisions English80.072.7
XNLI French French89.376.7
XNLI Arabic Arabic89.369.3
MASSIVE scenario Arabic89.563.0
MASSIVE intent Arabic91.584.0
OSACT4 offensive Arabic91.590.0
Tweet sentiment Arabic67.076.0
MIRACL rerank Arabic88.086.5
MIRACL pairwise Arabic91.090.5

Image datasets, accuracy

Dexy-Decision-9BJev 1.13
SST-2, as an image–94.5
Arabic sentiment, as an image–59.0
Oxford-IIIT Pet–83.5
VQAv2 yes/no–88.0
CountBench–94.5

Model pages: Dexy-Decision-9B · Jev 1.13. Latency for API models is the median per call through OpenRouter, timed from a cloud server with 8 calls in flight.