GPT-6 (High): benchmark results
Provider: OpenAI. Released 2026-09-03. Access: API.
Unified ELO 1797 ± 1, rank #7 of 1761 rated models, from 24 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| AA Omniscience | 43.73 | Score | 100 |
| AA MMMU-Pro | 86.42 | Accuracy (%) | 99.6 |
| AA GPQA Diamond | 94.95 | Accuracy (%) | 99.4 |
| ARC-AGI-1 | 98.5 | Accuracy (%) | 99.3 |
| Artificial Analysis Intelligence Index | 53.36 | Intelligence Index | 99 |
| AA Omniscience - Health | 53.7 | Accuracy (%) | 98.9 |
| AA Omniscience - Software Engineering (SWE) | 89.9 | Accuracy (%) | 98.8 |
| AA-Omniscience Accuracy | 61.13 | Accuracy (%) | 98.6 |
| AA Humanity's Last Exam | 53.06 | Accuracy (%) | 98.5 |
| ARC-AGI-2 | 92.08 | Accuracy (%) | 98.4 |
| AA Omniscience - Humanities & Social Sciences | 58.1 | Accuracy (%) | 98.2 |
| AA Omniscience - Business | 50.5 | Accuracy (%) | 98.1 |
Interactive version: theaggregate.ai/model?slug=gpt-6-high · How It Works · Data refreshed daily, snapshot 2026-09-05.