Head to head
GPT-6 Luna Decisions scores 76.0 and Jev 1.13 79.0 on the System One Index. GPT-6 Luna Decisions is more accurate on 5 of 16 text datasets, Jev 1.13 on 11. Same cases, same request, same scorer.
| GPT-6 Luna Decisions | Jev 1.13 | |
|---|---|---|
| System One Index | 76.0 | 79.0 |
| English | 75.2 | 78.5 |
| Arabic | 78.8 | 79.9 |
| Vision Index | 81.6 | 83.9 |
| Calibration error | 9.8% | 6.1% |
| Median latency | 215 ms | 150 ms |
| Cost per 1,000 calls | $0.040 | $0.028 |
| GPT-6 Luna Decisions | Jev 1.13 | |
|---|---|---|
| AG News English | 87.5 | 88.5 |
| SST-2 English | 93.5 | 94.5 |
| SST-5 English | 61.5 | 62.0 |
| DAIR Emotion English | 58.5 | 56.0 |
| Banking77 English | 77.5 | 79.0 |
| BoolQ English | 85.5 | 89.5 |
| XNLI English English | 66.7 | 86.0 |
| Typed decisions English | 71.0 | 72.7 |
| XNLI French French | 62.0 | 76.7 |
| XNLI Arabic Arabic | 54.0 | 69.3 |
| MASSIVE scenario Arabic | 66.0 | 63.0 |
| MASSIVE intent Arabic | 84.5 | 84.0 |
| OSACT4 offensive Arabic | 93.5 | 90.0 |
| Tweet sentiment Arabic | 80.0 | 76.0 |
| MIRACL rerank Arabic | 86.0 | 86.5 |
| MIRACL pairwise Arabic | 87.5 | 90.5 |
| GPT-6 Luna Decisions | Jev 1.13 | |
|---|---|---|
| SST-2, as an image | 88.5 | 94.5 |
| Arabic sentiment, as an image | 67.0 | 59.0 |
| Oxford-IIIT Pet | 87.0 | 83.5 |
| VQAv2 yes/no | 80.5 | 88.0 |
| CountBench | 85.0 | 94.5 |
Model pages: GPT-6 Luna Decisions ยท Jev 1.13. Latency for API models is the median per call through OpenRouter, timed from one machine.