TableVista - Simple Structure: leaderboard

Accuracy on the simple-grid tables (Web-style rendering) of TableVista (2026), a multimodal table reasoning benchmark in which models answer questions about rendered table images directly with thinking disabled, scored by exact match backed by a semantic-equivalence check. Percent; higher is better.

Metric: Accuracy (%) on the 300 simple-grid tables of the 3,000 TableVista questions, Web rendering, direct-output prompt with thinking disabled (reasoning effort none for the GPT models), exact match with a GPT-5-mini semantic-equivalence check on exact-match failures; higher is better. Source: arxiv.org. Saturation forecast: Around February 2027. 29 models tracked. Updated 7 Oct 2026.

Top models

#ModelSettingScoreOverall rank
1GPT-5.4Non-reasoning73#135
2Gemma 4 31B (IT)57.3#350
3Llama 4 Maverick Instruct FP855.3#1023
4Qwen 2.5 VL 72B Instruct52.3#841
5GPT-5.4 MiniNon-reasoning52#406
6Qwen 2.5 VL 32B Instruct51#1062
7Llama 4 Scout Instruct50.7#1579
8Qwen 3.5 27BNon-reasoning50.3#375
9Gemma 3 27B (IT)46.7#1324
10InternVL3.5-30B-A3B46.7#1267
11Gemma 4 26B A4B (IT)45.3#556
12Qwen 3.5 122B A10BNon-reasoning45#346
13Qwen 3 VL 30B A3B Instruct44.3#895
14InternVL3.5-14B44.3#1504
15Qwen 3.5 35B A3BNon-reasoning44#492

No result here: #1 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.

Interactive version: theaggregate.ai/benchmark?slug=tablevista-simple-structure · How It Works · Data refreshed daily, snapshot 2026-11-21.