MiniGPT-4-v1-7B: benchmark results
Provider: KAUST. Released 2023-04-20.
Aggregate Rating 1151 ± 47, rank #9795 of 10489 rated models, from 417 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| OpenVLM MMT-Bench - Tech Engineering | 33.3 | Score (%) | 62.4 |
| OpenVLM COCO Captions | 0.8 | CIDEr | 59.8 |
| OpenVLM MMT-Bench - Behavior Anomaly Detection | 40 | Score (%) | 58.3 |
| CoreCognition Image only Conservation | 44.44 | Accuracy (%) | 55.3 |
| OpenVLM MMT-Bench - Spot The Diff | 25 | Score (%) | 51.7 |
| CoreCognition Image only Perspective | 34.42 | Accuracy (%) | 48 |
| OpenVLM MME - Text Translation | 85 | Score | 40.6 |
| OpenVLM MMT-Bench - Science | 25 | Score (%) | 38.8 |
| OpenVLM MME - Numerical Calculation | 62.5 | Score | 38 |
| OpenVLM MMT-Bench - Next Image Prediction | 25 | Score (%) | 37.6 |
| OpenVLM MME - Code Reasoning | 70 | Score | 34.8 |
| OpenVLM MMT-Bench - Spot The Similarity | 55 | Score (%) | 33 |
Interactive version: theaggregate.ai/model?slug=minigpt-4-v1-7b · How It Works · Data refreshed daily, snapshot 2026-11-20.