Upstage · OpenRouter API
Solar Decide Flash by Upstage scores 63.6 on the System One Index (21st of 24). Calibration error 14.4%, median latency 586 ms.
| Dataset | Accuracy | Rank | Best on the board | Calibration error |
|---|---|---|---|---|
| AG News English | 86.0 | 19th of 24 | 92.5 | 10.2% |
| SST-2 English | 90.5 | 16th of 24 | 97.0 | 7.1% |
| SST-5 English | 44.5 | 19th of 24 | 62.5 | 31.5% |
| DAIR Emotion English | 54.5 | 20th of 24 | 91.5 | 32.2% |
| Banking77 English | 1.0 | 22nd of 24 | 98.5 | 0.3% |
| BoolQ English | 86.0 | 15th of 24 | 92.5 | 7.5% |
| XNLI English English | 65.3 | 23rd of 24 | 93.3 | 21.0% |
| Typed decisions English | 62.5 | 17th of 24 | 79.0 | 12.5% |
| XNLI French French | 61.3 | 23rd of 24 | 90.7 | 21.4% |
| XNLI Arabic Arabic | 52.0 | 25th of 26 | 89.3 | 27.9% |
| MASSIVE scenario Arabic | 46.0 | 22nd of 26 | 89.5 | 32.7% |
| MASSIVE intent Arabic | 63.0 | 20th of 26 | 91.5 | 18.3% |
| OSACT4 offensive Arabic | 84.5 | 15th of 26 | 95.5 | 8.2% |
| Tweet sentiment Arabic | 65.5 | 11th of 26 | 80.0 | 21.8% |
| MIRACL rerank Arabic | 71.5 | 18th of 26 | 90.5 | 5.7% |
| MIRACL pairwise Arabic | 83.0 | 19th of 26 | 91.0 | 7.2% |
upstage/solar-decide-flashRanks count the models charted on the leaderboard. Every model gets the same cases and the same request, under the benchmark rules.