InternVL3-9B-Instruct: benchmark results
Provider: Shanghai AI Lab. Access: Open.
Aggregate Rating 1486 ± 32, rank #2572 of 10916 rated models, from 31 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| DatBench - Spatial Reasoning | 32.9 | Accuracy (%) | 96.2 |
| FewMMBench (Zero-Shot) | 76.9 | Mean accuracy (%) on the FewMMBench query sets (about 250 mu | 78 |
| DatBench - Scene OCR | 66 | Accuracy (%) | 76.9 |
| DatBench - Diagrams & Tables | 40.3 | Accuracy (%) | 73.1 |
| DatBench - Counting | 91.4 | Accuracy (%) | 69.2 |
| FewMMBench (4-Shot, Similar + CoT) | 73.3 | Mean accuracy (%) on the FewMMBench query sets (about 250 mu | 68 |
| FewMMBench (8-Shot, Similar + CoT) | 72.5 | Mean accuracy (%) on the FewMMBench query sets (about 250 mu | 68 |
| MERA Multi - ruHHH-Video | 38.3 | Score (%) | 64.3 |
| FewMMBench (4-Shot, Random) | 75.6 | Mean accuracy (%) on the FewMMBench query sets (about 250 mu | 64 |
| FewMMBench (4-Shot, Similar) | 74.8 | Mean accuracy (%) on the FewMMBench query sets (about 250 mu | 64 |
| FewMMBench (8-Shot, Random) | 74.2 | Mean accuracy (%) on the FewMMBench query sets (about 250 mu | 64 |
| FewMMBench (8-Shot, Similar) | 74.3 | Mean accuracy (%) on the FewMMBench query sets (about 250 mu | 64 |
Interactive version: theaggregate.ai/model?slug=internvl3-9b-instruct · How It Works · Data refreshed daily, snapshot 2026-11-08.