DiningBench - Protein Estimation: leaderboard
DiningBench (2026) evaluates vision-language models on multi-view user photos of restaurant dishes, testing dish classification against same-menu distractors, nutrition estimation and food visual question answering. This board is the mean absolute percentage error of the estimated protein; lower is better.
Metric: Mean absolute percentage error (%) of the estimated protein over the 1,650 DiningBench nutrition-estimation dishes (multi-view restaurant photos; ground truth from merchant labels or ingredient-based estimates cross-checked against USDA FoodData Central); lower is better. Source: arxiv.org. Saturation forecast: Around May 2027. 29 models tracked. Updated 7 Oct 2026.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 3 Pro (Preview) | 23.88 | #119 |
| 2 | Gemini 3 Flash (Preview) | 25.23 | #152 |
| 3 | Qwen 3 VL 30B A3B Instruct | 32.01 | #895 |
| 4 | GPT-4.1 | 32.42 | #505 |
| 5 | Qwen 2.5 VL 32B Instruct | 32.73 | #1062 |
| 6 | O4 Mini | 32.96 | #368 |
| 7 | Claude Sonnet 4.5 | 33.15 | #261 |
| 8 | Qwen 2.5 VL 72B Instruct | 33.38 | #841 |
| 9 | GPT-5 | 33.4 | #178 |
| 10 | Qwen 3 VL 8B Instruct | 33.87 | #1029 |
| 11 | Qwen 2.5 VL 7B Instruct | 34.67 | #1902 |
| 12 | GPT-4o Mini | 35.08 | #1667 |
| 13 | InternVL3.5-38B | 35.59 | #1021 |
| 14 | Qwen 3 VL 4B Instruct | 36 | #1332 |
| 15 | GPT-4o | 36.03 | #751 |
No result here: #1 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.
Interactive version: theaggregate.ai/benchmark?slug=diningbench-protein-estimation · How It Works · Data refreshed daily, snapshot 2026-11-21.