InternVL2.5-26B-MPO: benchmark results
Provider: Shanghai AI Lab. Released 2024-12-20. Access: Open.
Aggregate Rating 1557 ± 39, rank #1600 of 10916 rated models, from 433 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| OpenVLM MMT-Bench - Jigsaw Puzzle Solving | 65 | Score (%) | 100 |
| OpenVLM POPE | 90.5 | Overall (%) | 99.8 |
| OpenVLM MMBench V1.1 EN - Function Reasoning | 97.8 | Accuracy (%) | 99.5 |
| OpenVLM MMT-Bench - Facial Expression Change Recognition | 100 | Score (%) | 99.5 |
| OpenVLM MMT-Bench - Google Apps | 75 | Score (%) | 99.3 |
| OpenVLM MMT-Bench - Meme Image Understanding | 100 | Score (%) | 99.3 |
| OpenVLM SEEDBench IMG - Instance Location | 74.5 | Accuracy (%) | 99.3 |
| OpenVLM SEEDBench IMG - Instance Attributes | 80.5 | Accuracy (%) | 99.1 |
| SEEDBench IMG | 78.3 | Accuracy (%) | 98.9 |
| OpenVLM MMT-Bench - I2IT | 55 | Score (%) | 98.8 |
| OpenVLM OCRBench - Doc-oriented VQA | 93 | Correct Answers (%) | 98.8 |
| OpenVLM CCBench - Sketch Reasoning | 95.6 | Accuracy (%) | 98.7 |
Interactive version: theaggregate.ai/model?slug=internvl2-5-26b-mpo · How It Works · Data refreshed daily, snapshot 2026-11-08.