TomaBench: leaderboard
TomaBench (Truong et al., 2026) asks vision-language models zero-shot multiple-choice questions about real tomato leaf images from the TomaMMU dataset. Score is the unweighted mean accuracy of seven tasks, from healthy or diseased screening to disease, pathogen and symptom identification, in percent; higher is better.
Metric: Accuracy (%; zero-shot multiple-choice questions on real tomato leaf images, unweighted mean of seven task accuracies; option matched against the answer key). Source: arxiv.org. Saturation forecast: Around February 2027. 9 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 2.0 Flash | 70.3 | #702 |
| 2 | LFM2.5-VL-1.6B | 61.3 | #3931 |
| 3 | Gemma 4 E4B | 24.8 | #2217 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=tomabench · How It Works · Data refreshed daily, snapshot 2026-11-12.