TableVision - Comparison: leaderboard
TableVision (2026) category of questions over images of hierarchical tables: L2 reasoning, comparing located values (113 questions). Exact-match accuracy (%), zero-shot; higher is better.
Metric: Exact-match accuracy (%) on L2 reasoning, comparing located values (113 questions), zero-shot, the model answers directly from the table image; TableVision test set; higher is better. Source: arxiv.org. Saturation forecast: Around February 2027. 10 models tracked. Updated 7 Oct 2026.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | HiPPO | 65.5 | |
| 2 | Qwen 3 VL 8B Instruct | 62.3 | #1029 |
| 3 | Ovis2.5-9B | 61.9 | #1225 |
| 4 | MiniCPM-V-2.6 | 58.4 | #3267 |
| 5 | Glyph | 58.4 | |
| 6 | GLM-4.1V-9B | 53.1 | #1125 |
| 7 | DeepSeek-VL2 | 36.3 | #2927 |
| 8 | InternVL3-8B | 27.4 | #1782 |
| 9 | LLaVA-OneVision-7B | 21.2 | #2981 |
| 10 | Phi-3.5-Vision Instruct | 10.6 | #6268 |
No result here: #1 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.
Interactive version: theaggregate.ai/benchmark?slug=tablevision-comparison · How It Works · Data refreshed daily, snapshot 2026-11-21.