~/system-one-bench / compare/nace-ai-drex-max-vs-patrick-18-dexy-decision-9b
Head to head
Drex Max scores 84.1 and Dexy-Decision-9B 84.7 on the System One Index. Drex Max is more accurate on 8 of 16 text datasets, Dexy-Decision-9B on 7, and 1 are tied. Same cases, same request, same scorer.
| Drex Max | Dexy-Decision-9B | |
|---|---|---|
| System One Index | 84.1 | 84.7 |
| English | 82.8 | 82.3 |
| Arabic | 84.5 | 86.8 |
| Vision Index | – | – |
| Calibration error | 6.2% | 3.5% |
| Median latency | 277 ms | – |
| Cost per 1,000 calls | $0.021 | – |
| Drex Max | Dexy-Decision-9B | |
|---|---|---|
| AG News English | 90.5 | 91.5 |
| SST-2 English | 96.0 | 95.5 |
| SST-5 English | 59.0 | 60.0 |
| DAIR Emotion English | 65.0 | 54.0 |
| Banking77 English | 90.5 | 93.0 |
| BoolQ English | 93.0 | 89.5 |
| XNLI English English | 95.3 | 94.7 |
| Typed decisions English | 73.0 | 80.0 |
| XNLI French French | 91.3 | 89.3 |
| XNLI Arabic Arabic | 89.3 | 89.3 |
| MASSIVE scenario Arabic | 73.0 | 89.5 |
| MASSIVE intent Arabic | 88.0 | 91.5 |
| OSACT4 offensive Arabic | 93.5 | 91.5 |
| Tweet sentiment Arabic | 69.0 | 67.0 |
| MIRACL rerank Arabic | 89.5 | 88.0 |
| MIRACL pairwise Arabic | 89.5 | 91.0 |
Model pages: Drex Max · Dexy-Decision-9B. Latency for API models is the median per call through OpenRouter, timed from a cloud server with 8 calls in flight.