~/system-one-bench / compare/celeris-1-decision-vs-microsoft-decision-1
Head to head
Celeris-1 Decision scores 83.6 and Microsoft-Decision-1 82.0 on the System One Index. Celeris-1 Decision is more accurate on 9 of 16 text datasets, Microsoft-Decision-1 on 6, and 1 are tied. Same cases, same request, same scorer.
| Celeris-1 Decision | Microsoft-Decision-1 | |
|---|---|---|
| System One Index | 83.6 | 82.0 |
| English | 83.2 | 81.0 |
| Arabic | 83.1 | 83.2 |
| Vision Index | โ | โ |
| Calibration error | 2.6% | 2.8% |
| Median latency | โ | 240 ms |
| Cost per 1,000 calls | $0.014 | $0.010 |
| Celeris-1 Decision | Microsoft-Decision-1 | |
|---|---|---|
| AG News English | 90.5 | 89.0 |
| SST-2 English | 95.5 | 95.5 |
| SST-5 English | 61.0 | 62.5 |
| DAIR Emotion English | 72.5 | 57.0 |
| Banking77 English | 89.0 | 87.0 |
| BoolQ English | 89.5 | 88.5 |
| XNLI English English | 91.3 | 89.3 |
| Typed decisions English | 76.4 | 79.0 |
| XNLI French French | 90.0 | 82.0 |
| XNLI Arabic Arabic | 80.0 | 72.0 |
| MASSIVE scenario Arabic | 73.5 | 83.5 |
| MASSIVE intent Arabic | 88.0 | 89.0 |
| OSACT4 offensive Arabic | 94.5 | 88.5 |
| Tweet sentiment Arabic | 69.0 | 68.5 |
| MIRACL rerank Arabic | 86.5 | 90.0 |
| MIRACL pairwise Arabic | 90.0 | 91.0 |
Model pages: Celeris-1 Decision ยท Microsoft-Decision-1. Latency for API models is the median per call through OpenRouter, timed from a cloud server with 8 calls in flight.