system-one-bench

~/system-one-bench / compare/patrick-18-dexy-decision-9b-vs-together-ai-tev1-4b-experimental

Head to head

Dexy-Decision-9B vs Tev1 4B Experimental

Dexy-Decision-9B scores 84.7 and Tev1 4B Experimental 71.5 on the System One Index. Dexy-Decision-9B is more accurate on 12 of 16 text datasets, Tev1 4B Experimental on 3, and 1 are tied. Same cases, same request, same scorer.

Overall

Dexy-Decision-9BTev1 4B Experimental
System One Index84.771.5
English82.366.4
Arabic86.875.8
Vision Index––
Calibration error3.5%5.3%
Median latency–232 ms
Cost per 1,000 calls–$0.019

Text datasets, accuracy

Dexy-Decision-9BTev1 4B Experimental
AG News English91.591.5
SST-2 English95.594.0
SST-5 English60.043.0
DAIR Emotion English54.056.5
Banking77 English93.01.0
BoolQ English89.591.0
XNLI English English94.787.3
Typed decisions English80.067.0
XNLI French French89.382.7
XNLI Arabic Arabic89.376.0
MASSIVE scenario Arabic89.556.5
MASSIVE intent Arabic91.570.0
OSACT4 offensive Arabic91.584.5
Tweet sentiment Arabic67.070.5
MIRACL rerank Arabic88.085.5
MIRACL pairwise Arabic91.087.5

Model pages: Dexy-Decision-9B · Tev1 4B Experimental. Latency for API models is the median per call through OpenRouter, timed from a cloud server with 8 calls in flight.