GPT-6 Sol (Non-reasoning): benchmark results
Provider: OpenAI. Released 2026-09-22. Access: API.
Unified ELO 1658 ± 1, rank #191 of 1906 rated models, from 63 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| AA-Omniscience Index - Software Engineering (SWE) - Rust | 66 | Omniscience Index | 90 |
| AA-Omniscience Index - Software Engineering (SWE) - Go | 50 | Omniscience Index | 87.8 |
| AA-Omniscience Index - Software Engineering (SWE) - Python | 47.5 | Omniscience Index | 87.4 |
| ParseBench - Tables | 89.49 | GTRM Score (GriTS + TableRecordMatch) | 87 |
| AA-Omniscience Accuracy | 45.18 | Accuracy (%) | 86.3 |
| AA-Omniscience Index - Software Engineering (SWE) - C | 60 | Omniscience Index | 85.6 |
| AA-Omniscience Index - Software Engineering (SWE) - TypeScript | 44.44 | Omniscience Index | 84.9 |
| AA-Omniscience Index - Software Engineering (SWE) | 34 | Omniscience Index | 83.6 |
| AA-Omniscience Index - Software Engineering (SWE) - Dart | 16 | Omniscience Index | 83.4 |
| Artificial Analysis Intelligence Index | 28.09 | Intelligence Index | 83.1 |
| AA-Omniscience Index - Software Engineering (SWE) - JavaScript | 38.18 | Omniscience Index | 82.9 |
| AA-Omniscience Index - Software Engineering (SWE) - Java | -2 | Omniscience Index | 81.8 |
Interactive version: theaggregate.ai/model?slug=gpt-6-sol-non-reasoning · How It Works · Data refreshed daily, snapshot 2026-09-24.