GPT-5.2 (2025-12-11) (Medium): benchmark results
Provider: OpenAI. Released 2025-12-11. Access: API.
Unified ELO 1738 ± 26, rank #259 of 2131 rated models, from 23 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Epoch AI - Algotune | 2.05 | Score | 100 |
| SWE-rebench | 59.25 | Resolved (%) | 90.6 |
| Chess Puzzles (Epoch AI) | 40 | Accuracy (%) | 88.5 |
| PLCC - Vocabulary | 86 | Accuracy (%) | 86.2 |
| OTIS Mock AIME 2024-25 | 93.89 | Accuracy (%) | 83 |
| PLCC - Geography | 94 | Accuracy (%) | 82.6 |
| Terminal-Bench 2.0 | 64.9 | Accuracy (%) | 82.1 |
| Context-Bench Skills | 77.55 | Task Completion (%) | 81 |
| Epoch AI - GPQA Diamond | 87.88 | Accuracy (%) | 79.9 |
| PLCC - Overall | 85 | Mean category accuracy (%) | 79.9 |
| PLCC - History | 90 | Accuracy (%) | 78.8 |
| PLCC - Grammar | 82 | Accuracy (%) | 77.5 |
Interactive version: theaggregate.ai/model?slug=gpt-5-2-2025-12-11-medium · How It Works · Data refreshed daily, snapshot 2026-10-09.