Humanity's Last Exam - Calibration Error: leaderboard
Metric: Calibration Error (%, lower is better). Source: agi.safe.ai. 10 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | GPT-5 | 50 |
| 2 | Grok 4 | 56.4 |
| 3 | Gemini 3 Pro | 57.2 |
| 4 | Claude Sonnet 4.5 | 65 |
| 5 | GPT-5 Mini | 65 |
| 6 | Gemini 2.5 Pro | 72 |
| 7 | DeepSeek R1 | 73 |
| 8 | Gemini 2.5 Flash | 80 |
| 9 | O1 | 83 |
| 10 | GPT-4o | 89 |
Interactive version: theaggregate.ai/benchmark?slug=humanity-s-last-exam-calibration-error · How It Works · Data refreshed daily, snapshot 2026-09-05.