system-one-bench

~/system-one-bench / compare/liquid-ai-d1-1-vs-openai-gpt-6-luna-decisions

Head to head

d1.1 vs GPT-6 Luna Decisions

d1.1 scores 82.3 and GPT-6 Luna Decisions 76.0 on the System One Index. d1.1 is more accurate on 13 of 16 text datasets, GPT-6 Luna Decisions on 3. Same cases, same request, same scorer.

Overall

d1.1GPT-6 Luna Decisions
System One Index82.376.0
English81.375.2
Arabic83.378.8
Vision Index85.881.6
Calibration error2.1%9.8%
Median latency–215 ms
Cost per 1,000 calls$0.005$0.040

Text datasets, accuracy

d1.1GPT-6 Luna Decisions
AG News English90.587.5
SST-2 English98.093.5
SST-5 English60.061.5
DAIR Emotion English60.558.5
Banking77 English87.077.5
BoolQ English92.085.5
XNLI English English90.066.7
Typed decisions English72.071.0
XNLI French French83.362.0
XNLI Arabic Arabic79.354.0
MASSIVE scenario Arabic73.066.0
MASSIVE intent Arabic88.084.5
OSACT4 offensive Arabic92.593.5
Tweet sentiment Arabic72.580.0
MIRACL rerank Arabic88.586.0
MIRACL pairwise Arabic89.587.5

Image datasets, accuracy

d1.1GPT-6 Luna Decisions
SST-2, as an image96.588.5
Arabic sentiment, as an image61.067.0
Oxford-IIIT Pet90.587.0
VQAv2 yes/no89.580.5
CountBench91.585.0

Model pages: d1.1 · GPT-6 Luna Decisions. Latency for API models is the median per call through OpenRouter, timed from a cloud server with 8 calls in flight.