DiningBench - Fat Estimation: leaderboard
Metric: Mean absolute percentage error (%) of the estimated fat 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 March 2027. 29 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Gemini 3 Pro (Preview) | 26.53 |
| 2 | Gemini 3 Flash (Preview) | 27.09 |
| 3 | GPT-5 | 33.49 |
| 4 | O4 Mini | 37.8 |
| 5 | Gemma 3 12B (IT) | 41.12 |
| 6 | Gemini 2.5 Pro | 42.19 |
| 7 | Qwen 3 VL 30B A3B Instruct | 43.37 |
| 8 | Qwen 3 VL 30B A3B (Thinking) | 43.66 |
| 9 | Qwen 3 VL 4B Instruct | 44.55 |
| 10 | Gemini 2.5 Flash | 44.56 |
| 11 | Qwen 3 VL 8B Instruct | 48.02 |
| 12 | GPT-4.1 | 49.29 |
| 13 | Qwen 2.5 VL 32B Instruct | 50.05 |
| 14 | Claude Sonnet 4.5 | 50.82 |
| 15 | Qwen 2.5 VL 72B Instruct | 51.27 |
Interactive version: theaggregate.ai/benchmark?slug=diningbench-fat-estimation · How It Works · Data refreshed daily, snapshot 2026-10-07.