Head to head
Clef Flash scores 80.3 and Laya Router 60.4 on the System One Index. Clef Flash is more accurate on 14 of 16 text datasets, Laya Router on 2. Same cases, same request, same scorer.
| Clef Flash | Laya Router | |
|---|---|---|
| System One Index | 80.3 | 60.4 |
| English | 80.2 | 64.7 |
| Arabic | 80.5 | 53.2 |
| Vision Index | 78.0 | – |
| Calibration error | 3.5% | 20.0% |
| Median latency | 267 ms | – |
| Cost per 1,000 calls | $0.044 | – |
| Clef Flash | Laya Router | |
|---|---|---|
| AG News English | 90.0 | 92.5 |
| SST-2 English | 95.0 | 88.5 |
| SST-5 English | 57.0 | 34.0 |
| DAIR Emotion English | 57.0 | 53.5 |
| Banking77 English | 98.5 | 44.0 |
| BoolQ English | 88.5 | 81.5 |
| XNLI English English | 84.7 | 87.3 |
| Typed decisions English | 70.5 | 36.4 |
| XNLI French French | 80.0 | 76.0 |
| XNLI Arabic Arabic | 68.7 | 68.0 |
| MASSIVE scenario Arabic | 89.5 | 39.0 |
| MASSIVE intent Arabic | 90.5 | 32.5 |
| OSACT4 offensive Arabic | 89.5 | 82.5 |
| Tweet sentiment Arabic | 47.0 | 37.0 |
| MIRACL rerank Arabic | 88.0 | 31.5 |
| MIRACL pairwise Arabic | 90.0 | 82.0 |
| Clef Flash | Laya Router | |
|---|---|---|
| SST-2, as an image | 95.5 | – |
| Arabic sentiment, as an image | 33.5 | – |
| Oxford-IIIT Pet | 85.0 | – |
| VQAv2 yes/no | 88.0 | – |
| CountBench | 88.0 | – |
Model pages: Clef Flash · Laya Router. Latency for API models is the median per call through OpenRouter, timed from one machine.