system-one-bench

~/system-one-bench / compare/baseten-elephant-1-vs-together-ai-tev1-4b-experimental

Head to head

Elephant 1 vs Tev1 4B Experimental

Elephant 1 scores 82.1 and Tev1 4B Experimental 71.5 on the System One Index. Elephant 1 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

Elephant 1Tev1 4B Experimental
System One Index82.171.5
English81.366.4
Arabic82.075.8
Vision Indexโ€“โ€“
Calibration error3.7%5.3%
Median latencyโ€“232 ms
Cost per 1,000 calls$0.013$0.019

Text datasets, accuracy

Elephant 1Tev1 4B Experimental
AG News English89.091.5
SST-2 English94.594.0
SST-5 English58.543.0
DAIR Emotion English54.556.5
Banking77 English94.51.0
BoolQ English91.091.0
XNLI English English94.087.3
Typed decisions English74.267.0
XNLI French French90.082.7
XNLI Arabic Arabic79.376.0
MASSIVE scenario Arabic69.056.5
MASSIVE intent Arabic89.570.0
OSACT4 offensive Arabic91.084.5
Tweet sentiment Arabic66.070.5
MIRACL rerank Arabic88.085.5
MIRACL pairwise Arabic91.087.5

Model pages: Elephant 1 ยท Tev1 4B Experimental. Latency for API models is the median per call through OpenRouter, timed from a cloud server with 8 calls in flight.