Head to head
Decider V1.2 scores 84.6 and Jev 1.14 79.3 on the System One Index. Decider V1.2 is more accurate on 13 of 16 text datasets, Jev 1.14 on 2, and 1 are tied. Same cases, same request, same scorer.
| Decider V1.2 | Jev 1.14 | |
|---|---|---|
| System One Index | 84.6 | 79.3 |
| English | 82.0 | 78.8 |
| Arabic | 86.9 | 80.3 |
| Vision Index | 83.8 | 85.1 |
| Calibration error | 0.7% | 6.0% |
| Median latency | 178 ms | 136 ms |
| Cost per 1,000 calls | $0.007 | $0.028 |
| Decider V1.2 | Jev 1.14 | |
|---|---|---|
| AG News English | 91.5 | 89.0 |
| SST-2 English | 95.5 | 95.0 |
| SST-5 English | 60.0 | 60.0 |
| DAIR Emotion English | 60.0 | 57.5 |
| Banking77 English | 88.0 | 78.5 |
| BoolQ English | 91.5 | 90.0 |
| XNLI English English | 94.0 | 86.0 |
| Typed decisions English | 75.1 | 74.2 |
| XNLI French French | 90.0 | 76.7 |
| XNLI Arabic Arabic | 90.0 | 71.3 |
| MASSIVE scenario Arabic | 79.0 | 65.0 |
| MASSIVE intent Arabic | 91.0 | 85.0 |
| OSACT4 offensive Arabic | 95.0 | 87.0 |
| Tweet sentiment Arabic | 75.5 | 76.0 |
| MIRACL rerank Arabic | 86.5 | 87.5 |
| MIRACL pairwise Arabic | 91.5 | 90.0 |
| Decider V1.2 | Jev 1.14 | |
|---|---|---|
| SST-2, as an image | 94.5 | 93.5 |
| Arabic sentiment, as an image | 57.0 | 63.5 |
| Oxford-IIIT Pet | 85.5 | 88.5 |
| VQAv2 yes/no | 86.5 | 86.5 |
| CountBench | 95.5 | 93.5 |
Model pages: Decider V1.2 ยท Jev 1.14. Latency for API models is the median per call through OpenRouter, timed from a cloud server with 8 calls in flight.