VILA1.5-40B: benchmark results
Provider: NVIDIA. Released 2024-05-09. Access: Open.
Aggregate Rating 1456 ± 23, rank #3518 of 10916 rated models, from 311 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| OpenVLM MMMU - Design | 90 | Accuracy (%) | 98.9 |
| OpenVLM MMMU - Literature | 93.3 | Accuracy (%) | 98.1 |
| OpenVLM SEEDBench IMG - Scene Understanding | 80.7 | Accuracy (%) | 96.7 |
| OpenVLM SEEDBench IMG - Visual Reasoning | 82.5 | Accuracy (%) | 95.8 |
| OpenVLM MMBench V1.1 CN - Action Recognition | 93.2 | Accuracy (%) | 95.4 |
| OpenVLM MMBench V1.1 CN - Future Prediction | 85.3 | Accuracy (%) | 94.9 |
| OpenVLM MME - Scene | 166.8 | Score | 94.2 |
| OpenVLM SEEDBench IMG - Instance Identity | 80.8 | Accuracy (%) | 94 |
| OpenVLM MME - Posters | 187.4 | Score | 92.9 |
| OpenVLM MMMU - Agriculture | 63.3 | Accuracy (%) | 92.2 |
| OpenVLM MMMU - History | 80 | Accuracy (%) | 89.8 |
| OpenVLM MMMU - Diagnostics and Laboratory Medicine | 50 | Accuracy (%) | 89.6 |
Interactive version: theaggregate.ai/model?slug=vila1-5-40b · How It Works · Data refreshed daily, snapshot 2026-11-08.