InternVL2-40B: benchmark results
Provider: Shanghai AI Lab. Released 2024-07-08. Access: Open.
Aggregate Rating 1510 ± 19, rank #2207 of 10916 rated models, from 476 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| OpenVLM MMT-Bench - Camouflage Object Detection | 90 | Score (%) | 100 |
| OpenVLM MMT-Bench - Furniture Keypoint Detection | 80 | Score (%) | 99.8 |
| OpenVLM MMT-Bench - Instance Captioning | 100 | Score (%) | 99.8 |
| OpenVLM A-Bench - Major Object Recognition | 96.2 | Accuracy (%) | 99.7 |
| OpenVLM MMT-Bench - Celebrity Recognition | 100 | Score (%) | 99.3 |
| OpenVLM MMT-Bench - Text-to-Image Retrieval | 80 | Score (%) | 99.3 |
| OpenVLM MMT-Bench - KD | 76 | Score (%) | 99 |
| OpenVLM MMBench V1.1 CN - Attribute Recognition | 97.8 | Accuracy (%) | 98.9 |
| OpenVLM CCBench - Sketch Reasoning | 95.6 | Accuracy (%) | 98.7 |
| OpenVLM MMBench V1.0 EN - RR | 89.1 | Accuracy (%) | 98.7 |
| OpenVLM MMBench V1.1 CN - Image Emotion | 87.8 | Accuracy (%) | 98.6 |
| OpenVLM MMT-Bench - Sports Recognition | 100 | Score (%) | 98.5 |
Interactive version: theaggregate.ai/model?slug=internvl2-40b · How It Works · Data refreshed daily, snapshot 2026-11-08.