InternVL3-2B: benchmark results
Provider: Shanghai AI Lab. Released 2025-04-11. Access: Open.
Aggregate Rating 1466 ± 11, rank #2997 of 10916 rated models, from 712 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| ProactiveBench - MVP-N | 20.1 | Trajectory accuracy (%) on the uninformative multi-view obje | 100 |
| RoadSafe365 - Weather | 75.63 | Four-option multiple-choice accuracy (%) on the weather ques | 100 |
| OpenVLM MMT-Bench - HLN | 80 | Score (%) | 99.8 |
| OpenVLM MMT-Bench - Order Hallucination | 65 | Score (%) | 99 |
| OpenVLM MMT-Bench - Musical Instrument Recognition | 100 | Score (%) | 98.3 |
| OpenVLM MMBench V1.1 CN - Social Relation | 91.2 | Accuracy (%) | 98.1 |
| OpenVLM MMT-Bench - MeViS | 55 | Score (%) | 98.1 |
| OpenVLM POPE | 90.1 | Overall (%) | 98.1 |
| OpenVLM MMBench V1.1 EN - Social Relation | 92.3 | Accuracy (%) | 97.7 |
| OpenVLM MMT-Bench - Multiple Image Captioning | 100 | Score (%) | 97.6 |
| ProactiveBench - VSOD | 31 | Trajectory accuracy (%) on the temporally occluded video fra | 97.6 |
| OpenVLM CCBench - Calligraphy and Painting | 89.5 | Accuracy (%) | 97.4 |
Interactive version: theaggregate.ai/model?slug=internvl3-2b · How It Works · Data refreshed daily, snapshot 2026-11-08.