TableVision - Retrieval: leaderboard
TableVision (2026) category of questions over images of hierarchical tables: L1 perception, locating and reading specific cells (371 questions). Exact-match accuracy (%), zero-shot; higher is better.
Metric: Exact-match accuracy (%) on L1 perception, locating and reading specific cells (371 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 | GLM-4.1V-9B | 80.6 | #1125 |
| 2 | Qwen 3 VL 8B Instruct | 78.8 | #1029 |
| 3 | Glyph | 78.4 | |
| 4 | Ovis2.5-9B | 76.5 | #1225 |
| 5 | HiPPO | 73.6 | |
| 6 | MiniCPM-V-2.6 | 73 | #3267 |
| 7 | DeepSeek-VL2 | 61.2 | #2927 |
| 8 | LLaVA-OneVision-7B | 57.7 | #2981 |
| 9 | InternVL3-8B | 36.4 | #1782 |
| 10 | Phi-3.5-Vision Instruct | 24.8 | #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-retrieval · How It Works · Data refreshed daily, snapshot 2026-11-21.