InternVL3.5-2B-Instruct: benchmark results
Provider: Shanghai AI Lab. Released 2025-08-26. Access: Open.
Aggregate Rating 1386 ± 38, rank #6661 of 10916 rated models, from 38 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| EgoPoint-Bench - Simulation Function & State | 55.14 | Accuracy (%) on the Function & State questions among the 1,9 | 90.9 |
| DatBench - Math & Logic | 16.6 | Accuracy (%) | 50 |
| DatBench - Chart Understanding | 48.7 | Accuracy (%) | 42.3 |
| DatBench - Scene OCR | 46.2 | Accuracy (%) | 42.3 |
| EgoPoint-Bench - Real-World Function & State | 60.47 | Accuracy (%) on the Function & State questions among the 1,1 | 36.4 |
| EgoPoint-Bench - Simulation OCR | 66.97 | Accuracy (%) on the OCR questions among the 1,929 held-out s | 36.4 |
| DatBench - Spatial Reasoning | 13.7 | Accuracy (%) | 30.8 |
| MERA Multi - ruMathVQA | 1.1 | Score (%) | 29.7 |
| EgoPoint-Bench - Simulation | 51.74 | Accuracy (%) over the 1,929 held-out simulation test questio | 27.3 |
| EgoPoint-Bench - Simulation Basic Perception | 51.97 | Accuracy (%) on the Basic Perception questions among the 1,9 | 27.3 |
| DatBench - Document Understanding | 24.5 | Accuracy (%) | 24 |
| MERA Multi - CommonVideoQA | 16.1 | Score (%) | 23.8 |
Interactive version: theaggregate.ai/model?slug=internvl3-5-2b-instruct · How It Works · Data refreshed daily, snapshot 2026-11-08.