Head to head
Decider Flash V1.2 scores 81.9 and d3 nano 78.0 on the System One Index. Decider Flash V1.2 is more accurate on 13 of 16 text datasets, d3 nano on 3. Same cases, same request, same scorer.
| Decider Flash V1.2 | d3 nano | |
|---|---|---|
| System One Index | 81.9 | 78.0 |
| English | 80.0 | 77.9 |
| Arabic | 83.3 | 76.9 |
| Vision Index | 77.5 | – |
| Calibration error | 1.8% | 7.0% |
| Median latency | 104 ms | – |
| Cost per 1,000 calls | $0.005 | – |
| Decider Flash V1.2 | d3 nano | |
|---|---|---|
| AG News English | 89.0 | 88.0 |
| SST-2 English | 95.0 | 89.5 |
| SST-5 English | 54.0 | 55.5 |
| DAIR Emotion English | 57.0 | 59.5 |
| Banking77 English | 90.5 | 93.0 |
| BoolQ English | 89.0 | 88.0 |
| XNLI English English | 92.7 | 87.3 |
| Typed decisions English | 73.0 | 62.6 |
| XNLI French French | 88.0 | 86.0 |
| XNLI Arabic Arabic | 85.3 | 82.0 |
| MASSIVE scenario Arabic | 70.5 | 61.5 |
| MASSIVE intent Arabic | 86.0 | 81.0 |
| OSACT4 offensive Arabic | 90.5 | 79.0 |
| Tweet sentiment Arabic | 71.5 | 71.0 |
| MIRACL rerank Arabic | 88.5 | 77.5 |
| MIRACL pairwise Arabic | 90.5 | 86.5 |
| Decider Flash V1.2 | d3 nano | |
|---|---|---|
| SST-2, as an image | 93.5 | – |
| Arabic sentiment, as an image | 35.0 | – |
| Oxford-IIIT Pet | 84.0 | – |
| VQAv2 yes/no | 87.5 | – |
| CountBench | 87.5 | – |
Model pages: Decider Flash V1.2 · d3 nano. Latency for API models is the median per call through OpenRouter, timed from a cloud server with 8 calls in flight.