InternVL3.5-8B: benchmark results
OpenGVLab's 8B InternVL3.5 vision-language model (August 2025) with a Qwen3 backbone and Cascade RL training for multimodal reasoning. Provider: Shanghai AI Lab. Released 2025-08-26. Access: Open.
Aggregate Rating 1554 ± 5, rank #1526 of 10800 rated models, from 1432 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| AgroVG - Multi-Target Set-F1 | 35.4 | Macro Set-F1 at IoU 0.5 on multi-target queries (%, times 10 | 100 |
| ArtECulture - Chinese | 54.34 | Accuracy (%; Chinese-speaking culture labels; zero-shot pred | 100 |
| CCTVBench | 11.38 | Quadruple accuracy (%): share of question pairs in which the | 100 |
| CCTVBench - Balanced Accuracy | 64.85 | Balanced accuracy (%) over all binary video-question instanc | 100 |
| Cambrian-S Release - MVBench † | 72.1 | Accuracy (%) | 100 |
| CapRiCorn-1K-V - Referential Consistency | 5.4 | Subject referential consistency (%; share of same-subject de | 100 |
| DriveSpatial - Understanding RMSE | 14.26 | Root mean squared error of numeric answers (distances, count | 100 |
| Holtercare-Bench - Diagnosis (Video) | 42.13 | Accuracy (%; multiple-choice closed QA on the overall diagno | 100 |
| Holtercare-Bench - Presence (Video) | 67.97 | Accuracy (%; multiple-choice closed QA on whether a given rh | 100 |
| HomeSafe-Bench | 28.42 | Weighted Safety Score (0-100) on the 438 hazardous household | 100 |
| MATEO (Image-only, Baseline) | 15 | Consistent temporal-execution-order accuracy (%) on MATEO's | 100 |
| MATEO (Image-only, Instructions) | 58 | Consistent temporal-execution-order accuracy (%) on MATEO's | 100 |
Interactive version: theaggregate.ai/model?slug=internvl3-5-8b · How It Works · Data refreshed daily, snapshot 2026-11-13.