Solar Pro 3 — benchmark results
Upstage's enterprise reasoning MoE (102B total/12B active), Korean/English bilingual (January 2026). Provider: Upstage. Released 2026-01-27. Access: API.
Unified ELO 1568 ± 36, rank #571 of 1776 rated models, from 43 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| AA IFBench | 71.2 | Accuracy (%) | 86.8 |
| AA TAU-2 Bench | 86.26 | Accuracy (%) | 78.9 |
| AA Omniscience - Software Engineering (SWE) - Dart | 30 | Accuracy (%) | 73.5 |
| AA Humanity's Last Exam | 10.15 | Accuracy (%) | 62.2 |
| AA Omniscience - Software Engineering (SWE) - Java | 19 | Accuracy (%) | 58 |
| AA Omniscience - Software Engineering (SWE) - Julia | 16 | Accuracy (%) | 57.8 |
| AA GPQA Diamond | 72.42 | Accuracy (%) | 56.7 |
| UGI - Natural Intelligence | 23.04 | NatInt Score | 55.1 |
| AA Omniscience - Software Engineering (SWE) - HTML | 32 | Accuracy (%) | 53.6 |
| AA Omniscience - Law | 11 | Accuracy (%) | 50.4 |
| AA Omniscience - Health | 19.1 | Accuracy (%) | 50.2 |
| AA Omniscience - Software Engineering (SWE) - TypeScript | 21.11 | Accuracy (%) | 48.3 |
Interactive version: theaggregate.ai/model?slug=solar-pro-3 · How the rankings work · Data refreshed daily, snapshot 2026-07-22.