SuperGPQA: leaderboard
Graduate-level QA benchmark spanning 285 disciplines. 10,000+ questions at graduate-level difficulty from verified experts. Tests deep, specialized knowledge beyond typical benchmarks.
Metric: Accuracy (%). Source: supergpqa.github.io. Saturation forecast: Around January 2027. 69 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 3 Pro (Preview) | 73.75 | #66 |
| 2 | GPT-5.2 Pro | 67.13 | #61 |
| 3 | GPT-5.1 (Thinking) | 66.35 | #135 (GPT-5.1) |
| 4 | GPT-5 | 65.63 | #98 |
| 5 | Gemini 2.5 Pro | 64.51 | #152 |
| 6 | Grok 4 | 62.99 | #181 |
| 7 | DeepSeek R1 | 61.82 | #251 |
| 8 | Claude Opus 4 (20250514) | 61.43 | #154 |
| 9 | GPT-5.2 (Thinking) | 61.25 | #110 (GPT-5.2) |
| 10 | O1 (2024-12-17) | 60.24 | #149 |
| 11 | DeepSeek V3.1 | 59.32 | #264 |
| 12 | Gemini 2.5 Flash | 58.91 | #245 |
| 13 | GPT-5 Mini | 58.82 | #186 |
| 14 | Kimi K2 (0711) | 58.08 | #492 |
| 15 | GPT-5 Chat | 57.36 | #241 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=supergpqa · How It Works · Data refreshed daily, snapshot 2026-10-20.