MMR-AD - MPDD Localization: leaderboard
Zero-shot industrial anomaly localization on the MPDD images of MMR-AD (2026): given a normal reference and a test image, the model draws a box around the anomaly, counted correct at an IoU of at least 0.1 with an annotated region. Accuracy in percent, averaged over product classes; higher is better.
Metric: Anomaly localization accuracy (%) on MPDD test images (from MMR-AD): a predicted anomaly box counts as correct at IoU of at least 0.1 with an annotated anomalous region, zero-shot, the model sees a spatially aligned normal reference image and the test image with the MMR-AD instruction template; dataset-level average over the subdataset's product classes; higher is better. Source: arxiv.org. Saturation forecast: Around April 2027. 13 models tracked. Updated 7 Oct 2026.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Qwen 2.5 VL 72B | 24.7 | #782 |
| 2 | O4 Mini | 23.9 | #372 |
| 3 | GPT-5 | 21.2 | #185 |
| 4 | QVQ-Max | 18 | #955 |
| 5 | Gemini 2.5 Pro | 16.3 | #280 |
| 6 | GPT-4o | 13.6 | #754 |
| 7 | Llama 4 Maverick | 10 | #1113 |
| 8 | Qwen 2.5 VL 7B | 7.2 | #1878 |
| 9 | InternVL3-38B | 5.4 | #948 |
| 10 | InternVL3.5-38B | 5.1 | #1026 |
| 11 | InternVL3.5-8B | 4.1 | #1618 |
| 12 | InternVL3-8B | 3.9 | #1795 |
| 13 | Gemma 3 27B | 3.5 | #1672 |
No result here: #2 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.
Interactive version: theaggregate.ai/benchmark?slug=mmr-ad-mpdd-localization · How It Works · Data refreshed daily, snapshot 2026-11-22.