~/system-one-bench / compare/liquid-ai-d1-vs-patrick-18-dexy-decision-9b
Head to head
d1 scores 82.7 and Dexy-Decision-9B 84.7 on the System One Index. d1 is more accurate on 5 of 16 text datasets, Dexy-Decision-9B on 9, and 2 are tied. Same cases, same request, same scorer.
| d1 | Dexy-Decision-9B | |
|---|---|---|
| System One Index | 82.7 | 84.7 |
| English | 81.3 | 82.3 |
| Arabic | 83.9 | 86.8 |
| Vision Index | – | – |
| Calibration error | 2.9% | 3.5% |
| Median latency | 188 ms | – |
| Cost per 1,000 calls | $0.015 | – |
| d1 | Dexy-Decision-9B | |
|---|---|---|
| AG News English | 90.0 | 91.5 |
| SST-2 English | 95.5 | 95.5 |
| SST-5 English | 56.5 | 60.0 |
| DAIR Emotion English | 63.5 | 54.0 |
| Banking77 English | 89.0 | 93.0 |
| BoolQ English | 90.0 | 89.5 |
| XNLI English English | 91.3 | 94.7 |
| Typed decisions English | 74.5 | 80.0 |
| XNLI French French | 84.7 | 89.3 |
| XNLI Arabic Arabic | 80.0 | 89.3 |
| MASSIVE scenario Arabic | 73.0 | 89.5 |
| MASSIVE intent Arabic | 90.0 | 91.5 |
| OSACT4 offensive Arabic | 95.5 | 91.5 |
| Tweet sentiment Arabic | 67.5 | 67.0 |
| MIRACL rerank Arabic | 90.5 | 88.0 |
| MIRACL pairwise Arabic | 91.0 | 91.0 |
Model pages: d1 · Dexy-Decision-9B. Latency for API models is the median per call through OpenRouter, timed from a cloud server with 8 calls in flight.