system-one-bench

~/system-one-bench / compare/celeris-1-decision-vs-cloudflare-clef-omni

Head to head

Celeris-1 Decision vs Clef Omni

Celeris-1 Decision scores 83.6 and Clef Omni 79.7 on the System One Index. Celeris-1 Decision is more accurate on 11 of 16 text datasets, Clef Omni on 5. Same cases, same request, same scorer.

Overall

Celeris-1 DecisionClef Omni
System One Index83.679.7
English83.278.6
Arabic83.180.5
Vision Index–72.2
Calibration error2.6%3.4%
Median latency–208 ms
Cost per 1,000 calls$0.014$0.074

Text datasets, accuracy

Celeris-1 DecisionClef Omni
AG News English90.589.0
SST-2 English95.597.0
SST-5 English61.051.0
DAIR Emotion English72.556.0
Banking77 English89.096.0
BoolQ English89.590.0
XNLI English English91.382.7
Typed decisions English76.467.0
XNLI French French90.082.7
XNLI Arabic Arabic80.073.3
MASSIVE scenario Arabic73.589.0
MASSIVE intent Arabic88.091.5
OSACT4 offensive Arabic94.590.5
Tweet sentiment Arabic69.050.0
MIRACL rerank Arabic86.583.0
MIRACL pairwise Arabic90.086.5

Image datasets, accuracy

Celeris-1 DecisionClef Omni
SST-2, as an image–89.5
Arabic sentiment, as an image–28.5
Oxford-IIIT Pet–88.0
VQAv2 yes/no–87.0
CountBench–68.0

Model pages: Celeris-1 Decision · Clef Omni. Latency for API models is the median per call through OpenRouter, timed from a cloud server with 8 calls in flight.