Head to head
Clef Omni scores 79.7 and Classify-1 30.6 on the System One Index. Clef Omni is more accurate on 15 of 16 text datasets, Classify-1 on 1. Same cases, same request, same scorer.
| Clef Omni | Classify-1 | |
|---|---|---|
| System One Index | 79.7 | 30.6 |
| English | 78.6 | 30.7 |
| Arabic | 80.5 | 30.6 |
| Vision Index | 72.2 | – |
| Calibration error | 3.4% | 3.2% |
| Median latency | 208 ms | 254 ms |
| Cost per 1,000 calls | $0.074 | $0.018 |
| Clef Omni | Classify-1 | |
|---|---|---|
| AG News English | 89.0 | 26.5 |
| SST-2 English | 97.0 | 50.0 |
| SST-5 English | 51.0 | 13.0 |
| DAIR Emotion English | 56.0 | 59.0 |
| Banking77 English | 96.0 | 1.0 |
| BoolQ English | 90.0 | 38.0 |
| XNLI English English | 82.7 | 30.7 |
| Typed decisions English | 67.0 | 27.2 |
| XNLI French French | 82.7 | 30.7 |
| XNLI Arabic Arabic | 73.3 | 30.7 |
| MASSIVE scenario Arabic | 89.0 | 3.0 |
| MASSIVE intent Arabic | 91.5 | 3.5 |
| OSACT4 offensive Arabic | 90.5 | 80.0 |
| Tweet sentiment Arabic | 50.0 | 35.0 |
| MIRACL rerank Arabic | 83.0 | 12.0 |
| MIRACL pairwise Arabic | 86.5 | 50.0 |
| Clef Omni | Classify-1 | |
|---|---|---|
| SST-2, as an image | 89.5 | – |
| Arabic sentiment, as an image | 28.5 | – |
| Oxford-IIIT Pet | 88.0 | – |
| VQAv2 yes/no | 87.0 | – |
| CountBench | 68.0 | – |
Model pages: Clef Omni · Classify-1. Latency for API models is the median per call through OpenRouter, timed from a cloud server with 8 calls in flight.