InternVL2.5-26B: benchmark results
Provider: Shanghai AI Lab. Released 2024-12-05. Access: Open.
Aggregate Rating 1502 ± 14, rank #2310 of 10916 rated models, from 309 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| TSHA | 80.8 | Final average (%): mean of the open-ended QA overall (GPT-4o | 100 |
| OpenVLM MMBench V1.1 CN - Attribute Recognition | 97.8 | Accuracy (%) | 98.9 |
| OpenVLM MMBench V1.1 CN - Action Recognition | 94 | Accuracy (%) | 97.5 |
| OpenVLM MMBench V1.1 CN - Object Localization | 89.5 | Accuracy (%) | 96.8 |
| OpenVLM MMBench V1.1 EN - Physical Relation | 74.7 | Accuracy (%) | 96.8 |
| OpenVLM MMBench V1.1 CN - CP | 84.3 | Accuracy (%) | 96.5 |
| OpenVLM MMBench V1.1 EN - Function Reasoning | 96.7 | Accuracy (%) | 96.3 |
| OpenVLM MMBench V1.1 CN - Image Emotion | 86.7 | Accuracy (%) | 95.8 |
| OpenVLM MMMU - Design | 86.7 | Accuracy (%) | 95.8 |
| OpenVLM MMBench V1.1 CN - Social Relation | 90.1 | Accuracy (%) | 95.2 |
| TSHA - Choice Questions | 88.3 | Choice-question overall (%): mean of yes/no hazard-presence | 95.2 |
| OpenVLM MMBench V1.1 EN - Image Style | 85.9 | Accuracy (%) | 95.1 |
Interactive version: theaggregate.ai/model?slug=internvl2-5-26b · How It Works · Data refreshed daily, snapshot 2026-11-08.