Head to head
Drex Max scores 84.1 and Decider Flash V1.2 81.9 on the System One Index. Drex Max is more accurate on 12 of 16 text datasets, Decider Flash V1.2 on 2, and 2 are tied. Same cases, same request, same scorer.
| Drex Max | Decider Flash V1.2 | |
|---|---|---|
| System One Index | 84.1 | 81.9 |
| English | 82.8 | 80.0 |
| Arabic | 84.5 | 83.3 |
| Vision Index | – | 77.5 |
| Calibration error | 6.2% | 1.8% |
| Median latency | 277 ms | 104 ms |
| Cost per 1,000 calls | $0.021 | $0.005 |
| Drex Max | Decider Flash V1.2 | |
|---|---|---|
| AG News English | 90.5 | 89.0 |
| SST-2 English | 96.0 | 95.0 |
| SST-5 English | 59.0 | 54.0 |
| DAIR Emotion English | 65.0 | 57.0 |
| Banking77 English | 90.5 | 90.5 |
| BoolQ English | 93.0 | 89.0 |
| XNLI English English | 95.3 | 92.7 |
| Typed decisions English | 73.0 | 73.0 |
| XNLI French French | 91.3 | 88.0 |
| XNLI Arabic Arabic | 89.3 | 85.3 |
| MASSIVE scenario Arabic | 73.0 | 70.5 |
| MASSIVE intent Arabic | 88.0 | 86.0 |
| OSACT4 offensive Arabic | 93.5 | 90.5 |
| Tweet sentiment Arabic | 69.0 | 71.5 |
| MIRACL rerank Arabic | 89.5 | 88.5 |
| MIRACL pairwise Arabic | 89.5 | 90.5 |
| Drex Max | Decider Flash V1.2 | |
|---|---|---|
| SST-2, as an image | – | 93.5 |
| Arabic sentiment, as an image | – | 35.0 |
| Oxford-IIIT Pet | – | 84.0 |
| VQAv2 yes/no | – | 87.5 |
| CountBench | – | 87.5 |
Model pages: Drex Max · Decider Flash V1.2. Latency for API models is the median per call through OpenRouter, timed from a cloud server with 8 calls in flight.