~/system-one-bench / compare/liquid-ai-d1-1-vs-microsoft-decision-1
Head to head
d1.1 scores 82.3 and Microsoft-Decision-1 82.0 on the System One Index. d1.1 is more accurate on 9 of 16 text datasets, Microsoft-Decision-1 on 6, and 1 are tied. Same cases, same request, same scorer.
| d1.1 | Microsoft-Decision-1 | |
|---|---|---|
| System One Index | 82.3 | 82.0 |
| English | 81.3 | 81.0 |
| Arabic | 83.3 | 83.2 |
| Vision Index | 85.8 | – |
| Calibration error | 2.1% | 2.8% |
| Median latency | – | 240 ms |
| Cost per 1,000 calls | $0.005 | $0.010 |
| d1.1 | Microsoft-Decision-1 | |
|---|---|---|
| AG News English | 90.5 | 89.0 |
| SST-2 English | 98.0 | 95.5 |
| SST-5 English | 60.0 | 62.5 |
| DAIR Emotion English | 60.5 | 57.0 |
| Banking77 English | 87.0 | 87.0 |
| BoolQ English | 92.0 | 88.5 |
| XNLI English English | 90.0 | 89.3 |
| Typed decisions English | 72.0 | 79.0 |
| XNLI French French | 83.3 | 82.0 |
| XNLI Arabic Arabic | 79.3 | 72.0 |
| MASSIVE scenario Arabic | 73.0 | 83.5 |
| MASSIVE intent Arabic | 88.0 | 89.0 |
| OSACT4 offensive Arabic | 92.5 | 88.5 |
| Tweet sentiment Arabic | 72.5 | 68.5 |
| MIRACL rerank Arabic | 88.5 | 90.0 |
| MIRACL pairwise Arabic | 89.5 | 91.0 |
| d1.1 | Microsoft-Decision-1 | |
|---|---|---|
| SST-2, as an image | 96.5 | – |
| Arabic sentiment, as an image | 61.0 | – |
| Oxford-IIIT Pet | 90.5 | – |
| VQAv2 yes/no | 89.5 | – |
| CountBench | 91.5 | – |
Model pages: d1.1 · Microsoft-Decision-1. Latency for API models is the median per call through OpenRouter, timed from a cloud server with 8 calls in flight.