WildTableBench - Value Modification: leaderboard
Metric: Accuracy (%) on the Value Modification subtype (C4-M), 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 January 2027. 21 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | GPT-5.2 (High) | 66.7 |
| 2 | Gemini 3 Pro | 61.9 |
| 3 | Gemini 3 Flash | 57.1 |
| 4 | Kimi K2.5 (Thinking) | 52.4 |
| 5 | Qwen 3 VL 235B A22B (Thinking) | 42.9 |
| 6 | Claude Sonnet 4.6 (High) | 42.9 |
| 7 | Seed 2.0 Pro (High) | 42.9 |
| 8 | Qwen 3 VL 32B Instruct | 33.3 |
| 9 | Claude Opus 4.6 (High) | 33.3 |
| 10 | GPT-5 Mini (High) | 28.6 |
| 11 | Qwen 3 VL 235B A22B Instruct | 23.8 |
| 12 | GLM-4.6V | 23.8 |
| 13 | O3 (High) | 9.5 |
| 14 | Qwen 3 VL 4B Instruct | 4.8 |
| 15 | Qwen 3 VL 32B (Thinking) | 4.8 |
Interactive version: theaggregate.ai/benchmark?slug=wildtablebench-value-modification · How It Works · Data refreshed daily, snapshot 2026-10-07.