InternLM-XComposer2-4KHD: benchmark results
Provider: Shanghai AI Lab. Released 2024-04-09. Access: Open.
Aggregate Rating 1393 ± 33, rank #6402 of 10916 rated models, from 425 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| OpenVLM MME - Artwork | 187.2 | Score | 99.8 |
| OpenVLM MMT-Bench - One Shot Detection | 95 | Score (%) | 99 |
| OpenVLM MMT-Bench - Image Season Recognition | 90 | Score (%) | 98.3 |
| OpenVLM MMT-Bench - Scene Graph Recognition | 95 | Score (%) | 95.1 |
| OpenVLM MMT-Bench - Table Structure Recognition | 100 | Score (%) | 94.7 |
| OpenVLM MMT-Bench - Weather Recognition | 100 | Score (%) | 94.4 |
| OpenVLM MMT-Bench - Action Quality Assessment | 35 | Score (%) | 94.2 |
| OpenVLM MMT-Bench - Building Recognition | 90 | Score (%) | 93.7 |
| OpenVLM MMT-Bench - Facial Expression Change Recognition | 95 | Score (%) | 93 |
| GeoMMBench - Spatial | 59.5 | Micro-averaged accuracy (%) on the spatial relation question | 92.6 |
| OpenVLM MMT-Bench - Traffic Participants Understanding | 65 | Score (%) | 92.5 |
| OpenVLM MMT-Bench - Polygon Localization | 65 | Score (%) | 91.7 |
Interactive version: theaggregate.ai/model?slug=internlm-xcomposer2-4khd · How It Works · Data refreshed daily, snapshot 2026-11-08.