Head to head
Solar Decide scores 68.8 and d3 nano 78.0 on the System One Index. Solar Decide is more accurate on 5 of 16 text datasets, d3 nano on 11. Same cases, same request, same scorer.
| Solar Decide | d3 nano | |
|---|---|---|
| System One Index | 68.8 | 78.0 |
| English | 64.7 | 77.9 |
| Arabic | 73.7 | 76.9 |
| Vision Index | – | – |
| Calibration error | 20.3% | 7.0% |
| Median latency | 715 ms | – |
| Cost per 1,000 calls | $0.042 | – |
| Solar Decide | d3 nano | |
|---|---|---|
| AG News English | 86.0 | 88.0 |
| SST-2 English | 91.5 | 89.5 |
| SST-5 English | 53.5 | 55.5 |
| DAIR Emotion English | 54.0 | 59.5 |
| Banking77 English | 1.0 | 93.0 |
| BoolQ English | 87.0 | 88.0 |
| XNLI English English | 74.7 | 87.3 |
| Typed decisions English | 69.9 | 62.6 |
| XNLI French French | 67.3 | 86.0 |
| XNLI Arabic Arabic | 58.0 | 82.0 |
| MASSIVE scenario Arabic | 55.5 | 61.5 |
| MASSIVE intent Arabic | 75.0 | 81.0 |
| OSACT4 offensive Arabic | 86.0 | 79.0 |
| Tweet sentiment Arabic | 71.5 | 71.0 |
| MIRACL rerank Arabic | 84.5 | 77.5 |
| MIRACL pairwise Arabic | 85.5 | 86.5 |
Model pages: Solar Decide · d3 nano. Latency for API models is the median per call through OpenRouter, timed from one machine.