system-one-bench

~/system-one-bench / compare/liquid-ai-d1-vs-meragpt-decider-1

Head to head

d1 vs Decider 1

d1 scores 82.7 and Decider 1 61.8 on the System One Index. d1 is more accurate on 15 of 16 text datasets, Decider 1 on 1. Same cases, same request, same scorer.

Overall

d1Decider 1
System One Index82.761.8
English81.367.4
Arabic83.954.7
Vision Indexโ€“โ€“
Calibration error2.9%7.9%
Median latency188 msโ€“
Cost per 1,000 calls$0.015$0.012

Text datasets, accuracy

d1Decider 1
AG News English90.087.5
SST-2 English95.593.0
SST-5 English56.555.5
DAIR Emotion English63.567.5
Banking77 English89.01.0
BoolQ English90.080.0
XNLI English English91.380.7
Typed decisions English74.574.2
XNLI French French84.767.3
XNLI Arabic Arabic80.067.3
MASSIVE scenario Arabic73.03.0
MASSIVE intent Arabic90.03.5
OSACT4 offensive Arabic95.586.0
Tweet sentiment Arabic67.564.0
MIRACL rerank Arabic90.581.0
MIRACL pairwise Arabic91.078.0

Model pages: d1 ยท Decider 1. Latency for API models is the median per call through OpenRouter, timed from a cloud server with 8 calls in flight.