system-one-bench

~/system-one-bench / compare/fastino-glide-vs-perplexity-decider-v1-2

Head to head

GLiDE vs Decider V1.2

GLiDE scores 84.1 and Decider V1.2 84.6 on the System One Index. GLiDE is more accurate on 8 of 16 text datasets, Decider V1.2 on 6, and 2 are tied. Same cases, same request, same scorer.

Overall

GLiDEDecider V1.2
System One Index84.184.6
English83.082.0
Arabic84.886.9
Vision Index82.583.8
Calibration error3.6%0.7%
Median latency–178 ms
Cost per 1,000 calls$0.017$0.007

Text datasets, accuracy

GLiDEDecider V1.2
AG News English92.091.5
SST-2 English96.095.5
SST-5 English61.560.0
DAIR Emotion English61.560.0
Banking77 English91.588.0
BoolQ English92.091.5
XNLI English English94.794.0
Typed decisions English75.075.1
XNLI French French88.090.0
XNLI Arabic Arabic83.390.0
MASSIVE scenario Arabic73.579.0
MASSIVE intent Arabic91.091.0
OSACT4 offensive Arabic95.095.0
Tweet sentiment Arabic73.575.5
MIRACL rerank Arabic87.086.5
MIRACL pairwise Arabic90.591.5

Image datasets, accuracy

GLiDEDecider V1.2
SST-2, as an image94.094.5
Arabic sentiment, as an image51.557.0
Oxford-IIIT Pet84.585.5
VQAv2 yes/no89.086.5
CountBench93.595.5

Model pages: GLiDE · Decider V1.2. Latency for API models is the median per call through OpenRouter, timed from a cloud server with 8 calls in flight.