system-one-bench

~/system-one-bench / compare/meragpt-decider-1-vs-vllm-semantic-router-d3-edge

Head to head

Decider 1 vs d3 edge

Decider 1 scores 61.8 and d3 edge 65.9 on the System One Index. Decider 1 is more accurate on 12 of 16 text datasets, d3 edge on 4. Same cases, same request, same scorer.

Overall

Decider 1d3 edge
System One Index61.865.9
English67.469.1
Arabic54.762.2
Vision Index––
Calibration error7.9%18.7%
Median latency––
Cost per 1,000 calls$0.012–

Text datasets, accuracy

Decider 1d3 edge
AG News English87.580.0
SST-2 English93.080.5
SST-5 English55.539.5
DAIR Emotion English67.557.0
Banking77 English1.090.0
BoolQ English80.079.5
XNLI English English80.784.0
Typed decisions English74.242.3
XNLI French French67.366.0
XNLI Arabic Arabic67.359.3
MASSIVE scenario Arabic3.041.0
MASSIVE intent Arabic3.555.0
OSACT4 offensive Arabic86.072.5
Tweet sentiment Arabic64.061.5
MIRACL rerank Arabic81.070.0
MIRACL pairwise Arabic78.076.0

Model pages: Decider 1 · d3 edge. Latency for API models is the median per call through OpenRouter, timed from a cloud server with 8 calls in flight.