Head to head
Clef Omni scores 79.7 and Kev 0.8B 63.6 on the System One Index. Clef Omni is more accurate on 13 of 16 text datasets, Kev 0.8B on 2, and 1 are tied. Same cases, same request, same scorer.
| Clef Omni | Kev 0.8B | |
|---|---|---|
| System One Index | 79.7 | 63.6 |
| English | 78.6 | 71.0 |
| Arabic | 80.5 | 54.9 |
| Vision Index | 72.2 | – |
| Calibration error | 3.4% | 4.3% |
| Median latency | 208 ms | – |
| Cost per 1,000 calls | $0.074 | – |
| Clef Omni | Kev 0.8B | |
|---|---|---|
| AG News English | 89.0 | 91.0 |
| SST-2 English | 97.0 | 90.0 |
| SST-5 English | 51.0 | 47.0 |
| DAIR Emotion English | 56.0 | 56.0 |
| Banking77 English | 96.0 | 80.5 |
| BoolQ English | 90.0 | 76.0 |
| XNLI English English | 82.7 | 81.3 |
| Typed decisions English | 67.0 | 46.3 |
| XNLI French French | 82.7 | 66.0 |
| XNLI Arabic Arabic | 73.3 | 58.7 |
| MASSIVE scenario Arabic | 89.0 | 38.5 |
| MASSIVE intent Arabic | 91.5 | 51.5 |
| OSACT4 offensive Arabic | 90.5 | 49.0 |
| Tweet sentiment Arabic | 50.0 | 57.0 |
| MIRACL rerank Arabic | 83.0 | 62.5 |
| MIRACL pairwise Arabic | 86.5 | 67.0 |
| Clef Omni | Kev 0.8B | |
|---|---|---|
| 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 · Kev 0.8B. Latency for API models is the median per call through OpenRouter, timed from one machine.