XpertBench - Humanities and Social Sciences: leaderboard
Metric: Weighted rubric score (%) on the XpertBench-Gold subset (245 expert-written open-ended professional tasks, 15-40 weighted binary checkpoints each); ShotJudge: Gemini 2.5 Pro judges every checkpoint with an expert-graded GPT-5 response as one-shot exemplar; humanities and social sciences tasks; higher is better. Source: arxiv.org. Saturation forecast: Around January 2027. 14 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Claude Opus 4.6 (Thinking) | 83.02 |
| 2 | GPT-5.4 (High) | 80.58 |
| 3 | GPT-5.2 (High) | 76.15 |
| 4 | Gemini 3 Pro | 71.45 |
| 5 | Gemini 3 Flash | 70.66 |
| 6 | Gemini 3.1 Pro (Preview) | 69.04 |
| 7 | GPT-5 (High) | 68.5 |
| 8 | Kimi K2 (Thinking) | 66.4 |
| 9 | DeepSeek V3.1 Terminus (Thinking) | 55.2 |
| 10 | Claude Sonnet 4.5 (Thinking) | 54.9 |
| 11 | Gemini 2.5 Pro | 52.3 |
| 12 | Doubao-Seed-1.6 (Thinking) | 50.6 |
| 13 | GLM-4.6 | 49.7 |
Interactive version: theaggregate.ai/benchmark?slug=xpertbench-humanities-and-social-sciences · How It Works · Data refreshed daily, snapshot 2026-10-07.