WildTableBench - Hypothetical: leaderboard
Accuracy in percent on the hypothetical category: results recomputed after removing rows, editing values or assuming a what-if condition, free-form questions about real-world table images collected from the web (WildTableBench, 2026), answers judged against the reference by GPT-5.2. Higher is better.
Metric: Accuracy (%) on the Hypothetical question category (C4, question-weighted over its subtypes), 928 human-verified questions over 402 real-world table images, free-form answers judged by GPT-5.2 against the reference answer, high reasoning effort where configurable; higher is better. Source: arxiv.org. Saturation forecast: Around February 2027. 21 models tracked. Updated 7 Oct 2026.
Top models
| # | Model | Setting | Score | Overall rank |
|---|---|---|---|---|
| 1 | Gemini 3 Pro | 75.3 | #146 | |
| 2 | Gemini 3 Flash | 55.1 | #180 | |
| 3 | Kimi K2.5 | Thinking | 55.1 | #272 |
| 4 | GPT-5.2 | High | 55.1 | #190 |
| 5 | Claude Opus 4.6 | High | 53.9 | #114 |
| 6 | Claude Sonnet 4.6 | High | 49.4 | #179 |
| 7 | Seed 2.0 Pro | High | 44.9 | #226 |
| 8 | Qwen 3 VL 235B A22B (Thinking) | 39.3 | #462 | |
| 9 | GPT-5 Mini | High | 33.7 | #373 |
| 10 | Qwen 3 VL 32B Instruct | 27 | #607 | |
| 11 | GLM-4.6V | 27 | #721 | |
| 12 | Qwen 3 VL 235B A22B Instruct | 21.3 | #541 | |
| 13 | Qwen 3 VL 32B (Thinking) | 20.2 | #577 | |
| 14 | Qwen 3 VL 8B (Thinking) | 16.9 | #924 | |
| 15 | O3 | High | 10.1 | #242 |
No result here: #2 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.
Interactive version: theaggregate.ai/benchmark?slug=wildtablebench-hypothetical · How It Works · Data refreshed daily, snapshot 2026-11-22.