system-one-bench

~/system-one-bench / compare/fastino-glide-vs-liquid-ai-d1

Head to head

GLiDE vs d1

GLiDE scores 84.1 and d1 82.7 on the System One Index. GLiDE is more accurate on 12 of 16 text datasets, d1 on 4. Same cases, same request, same scorer.

Overall

GLiDEd1
System One Index84.182.7
English83.081.3
Arabic84.883.9
Vision Index82.5–
Calibration error3.6%2.9%
Median latency–188 ms
Cost per 1,000 calls$0.017$0.015

Text datasets, accuracy

GLiDEd1
AG News English92.090.0
SST-2 English96.095.5
SST-5 English61.556.5
DAIR Emotion English61.563.5
Banking77 English91.589.0
BoolQ English92.090.0
XNLI English English94.791.3
Typed decisions English75.074.5
XNLI French French88.084.7
XNLI Arabic Arabic83.380.0
MASSIVE scenario Arabic73.573.0
MASSIVE intent Arabic91.090.0
OSACT4 offensive Arabic95.095.5
Tweet sentiment Arabic73.567.5
MIRACL rerank Arabic87.090.5
MIRACL pairwise Arabic90.591.0

Image datasets, accuracy

GLiDEd1
SST-2, as an image94.0–
Arabic sentiment, as an image51.5–
Oxford-IIIT Pet84.5–
VQAv2 yes/no89.0–
CountBench93.5–

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