GPT-5.2 (2025-12-11) (High): benchmark results
Provider: OpenAI. Released 2025-12-11. Access: API.
Unified ELO 1743 ± 15, rank #249 of 2131 rated models, from 67 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| OpenCompass Reasoning - Academic (CompassBench 2601) | 50.5 | Score (%) | 100 |
| OpenCompass Math - Competition (CompassBench 2601) | 70 | Score (%) | 98 |
| OpenCompass LLM - Reasoning (CompassBench 2601) | 60.1 | Score (%) | 96 |
| OpenCompass Reasoning - Common (CompassBench 2601) | 69.6 | Score (%) | 96 |
| Context-Bench Skills | 84.47 | Task Completion (%) | 95.2 |
| EsoBench | 26.87 | Score | 94.6 |
| OpenCompass Code - Competition (CompassBench 2601) | 62.9 | Score (%) | 92 |
| OpenCompass Knowledge - Science (CompassBench 2601) | 97 | Score (%) | 92 |
| OpenCompass LLM - Code (CompassBench 2601) | 71.2 | Score (%) | 92 |
| OpenCompass Code - Comprehensive (CompassBench 2601) | 79.5 | Score (%) | 90 |
| OTIS Mock AIME 2024-25 | 96.11 | Accuracy (%) | 88.7 |
| Chess Puzzles (Epoch AI) | 40 | Accuracy (%) | 88.5 |
Interactive version: theaggregate.ai/model?slug=gpt-5-2-2025-12-11-high · How It Works · Data refreshed daily, snapshot 2026-10-09.