Head to head
Kev 0.8B scores 63.6 and Decider V1.2 84.6 on the System One Index. Kev 0.8B is more accurate on 0 of 16 text datasets, Decider V1.2 on 16. Same cases, same request, same scorer.
| Kev 0.8B | Decider V1.2 | |
|---|---|---|
| System One Index | 63.6 | 84.6 |
| English | 71.0 | 82.0 |
| Arabic | 54.9 | 86.9 |
| Vision Index | – | 83.8 |
| Calibration error | 4.3% | 0.7% |
| Median latency | – | 178 ms |
| Cost per 1,000 calls | – | $0.007 |
| Kev 0.8B | Decider V1.2 | |
|---|---|---|
| AG News English | 91.0 | 91.5 |
| SST-2 English | 90.0 | 95.5 |
| SST-5 English | 47.0 | 60.0 |
| DAIR Emotion English | 56.0 | 60.0 |
| Banking77 English | 80.5 | 88.0 |
| BoolQ English | 76.0 | 91.5 |
| XNLI English English | 81.3 | 94.0 |
| Typed decisions English | 46.3 | 75.1 |
| XNLI French French | 66.0 | 90.0 |
| XNLI Arabic Arabic | 58.7 | 90.0 |
| MASSIVE scenario Arabic | 38.5 | 79.0 |
| MASSIVE intent Arabic | 51.5 | 91.0 |
| OSACT4 offensive Arabic | 49.0 | 95.0 |
| Tweet sentiment Arabic | 57.0 | 75.5 |
| MIRACL rerank Arabic | 62.5 | 86.5 |
| MIRACL pairwise Arabic | 67.0 | 91.5 |
| Kev 0.8B | Decider V1.2 | |
|---|---|---|
| SST-2, as an image | – | 94.5 |
| Arabic sentiment, as an image | – | 57.0 |
| Oxford-IIIT Pet | – | 85.5 |
| VQAv2 yes/no | – | 86.5 |
| CountBench | – | 95.5 |
Model pages: Kev 0.8B · Decider V1.2. Latency for API models is the median per call through OpenRouter, timed from a cloud server with 8 calls in flight.