Head to head
Microsoft-Decision-1 scores 82.0 and Tev1 4B Experimental 71.5 on the System One Index. Microsoft-Decision-1 is more accurate on 11 of 16 text datasets, Tev1 4B Experimental on 5. Same cases, same request, same scorer.
| Microsoft-Decision-1 | Tev1 4B Experimental | |
|---|---|---|
| System One Index | 82.0 | 71.5 |
| English | 81.0 | 66.4 |
| Arabic | 83.2 | 75.8 |
| Vision Index | – | – |
| Calibration error | 2.8% | 5.3% |
| Median latency | 240 ms | 232 ms |
| Cost per 1,000 calls | $0.010 | $0.019 |
| Microsoft-Decision-1 | Tev1 4B Experimental | |
|---|---|---|
| AG News English | 89.0 | 91.5 |
| SST-2 English | 95.5 | 94.0 |
| SST-5 English | 62.5 | 43.0 |
| DAIR Emotion English | 57.0 | 56.5 |
| Banking77 English | 87.0 | 1.0 |
| BoolQ English | 88.5 | 91.0 |
| XNLI English English | 89.3 | 87.3 |
| Typed decisions English | 79.0 | 67.0 |
| XNLI French French | 82.0 | 82.7 |
| XNLI Arabic Arabic | 72.0 | 76.0 |
| MASSIVE scenario Arabic | 83.5 | 56.5 |
| MASSIVE intent Arabic | 89.0 | 70.0 |
| OSACT4 offensive Arabic | 88.5 | 84.5 |
| Tweet sentiment Arabic | 68.5 | 70.5 |
| MIRACL rerank Arabic | 90.0 | 85.5 |
| MIRACL pairwise Arabic | 91.0 | 87.5 |
Model pages: Microsoft-Decision-1 · Tev1 4B Experimental. Latency for API models is the median per call through OpenRouter, timed from one machine.