InternVL3.5-4B-Instruct: benchmark results
Provider: Shanghai AI Lab. Released 2025-08-26. Access: Open.
Aggregate Rating 1505 ± 28, rank #2269 of 10916 rated models, from 42 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| SiT-Bench - Global Perception & Mapping | 25.79 | Accuracy (%; item-weighted over the category's subtasks) | 86.7 |
| SiT-Bench - Logic & Anomaly Detection | 38.44 | Accuracy (%; item-weighted over the category's subtasks) | 83.3 |
| DatBench - Math & Logic | 19.7 | Accuracy (%) | 69.2 |
| DatBench - Spatial Reasoning | 21.4 | Accuracy (%) | 69.2 |
| DatBench - Diagrams & Tables | 39.2 | Accuracy (%) | 65.4 |
| SiT-Bench - Navigation & Planning | 47.44 | Accuracy (%; item-weighted over the category's subtasks) | 63.3 |
| SiT-Bench | 39.95 | Accuracy (%; item-weighted over the 17 subtasks) | 62.1 |
| DatBench - Chart Understanding | 58.8 | Accuracy (%) | 61.5 |
| DatBench - Counting | 90.6 | Accuracy (%) | 57.7 |
| MCSBench - L3 | 22.9 | Accuracy (%) | 53.3 |
| MERA Multi - ruMathVQA | 2.6 | Score (%) | 51.7 |
| MCSBench - L2 | 50.11 | Accuracy (%) | 50.7 |
Interactive version: theaggregate.ai/model?slug=internvl3-5-4b-instruct · How It Works · Data refreshed daily, snapshot 2026-11-08.