GPT-5.6 Terra (Low): benchmark results
GPT-5.6 Terra evaluated at the low reasoning-effort setting. Provider: OpenAI. Released 2026-07-09. Access: API.
Unified ELO 1662 ± 1, rank #206 of 1761 rated models, from 30 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| AA Omniscience - Software Engineering (SWE) | 70.4 | Accuracy (%) | 91.8 |
| AA Terminal-Bench Hard | 43.94 | Accuracy (%) | 91.3 |
| AA Omniscience - Health | 41.1 | Accuracy (%) | 88.3 |
| AA-Omniscience Accuracy | 43.73 | Accuracy (%) | 86.7 |
| AA Omniscience - Humanities & Social Sciences | 40.3 | Accuracy (%) | 85.9 |
| AA CritPt | 9.43 | Accuracy (%) | 84.9 |
| AA Omniscience - Business | 33.9 | Accuracy (%) | 84.2 |
| Artificial Analysis Intelligence Index | 32.4 | Intelligence Index | 84 |
| AA Omniscience - Law | 34.3 | Accuracy (%) | 83.7 |
| AA Omniscience - Science, Engineering & Mathematics | 42.4 | Accuracy (%) | 82.1 |
| AA Humanity's Last Exam | 29.19 | Accuracy (%) | 81.2 |
| OTIS Mock AIME 2024-25 | 88.89 | Accuracy (%) | 77.6 |
Interactive version: theaggregate.ai/model?slug=gpt-5-6-terra-low · How It Works · Data refreshed daily, snapshot 2026-09-05.