Visual Document Retrieval on ViDoRe
87.3ArxivQ ScoreNomic-Embed-Multimodal
Evaluation Results
| Method | Links | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Nomic-Embed-Multimodal#Params=3B, Latency=–2026.06 | 87.3 | 59.6 | 89.8 | 69.4 | 97.5 | 93.3 | 95.7 | 97.9 | 86.3 | |
| ColQwen2.5#Params=2B, Latency=466.62026.06 | 87.2 | 63.3 | 91.3 | 79.5 | 98.6 | 96.3 | 96.3 | 98 | 88.8 | |
| LightSTAR#Params=2B, Latency=123.92026.06 | 86.2 | 63.5 | 91.7 | 81 | 99.5 | 95.4 | 95.8 | 99.8 | 89.1 | |
| VisRAG-Ret#Params=3B, Latency=1219.92026.06 | 80.6 | 42.7 | 85 | 49.3 | 94.4 | 92 | 87.7 | 92.7 | 78 | |
| ColPali#Params=3B, Latency=257.32026.06 | 79.1 | 54.4 | 81.8 | 65.8 | 96.2 | 91 | 92.7 | 94.4 | 81.9 | |
| SigLIP#Params=883M, Latency=–2026.06 | 50.2 | 31.3 | 69.7 | 27.5 | 67.8 | 73.5 | 75.3 | 83.1 | 59.8 | |
| E5-V#Params=8B, Latency=–2026.06 | 48.3 | 34.7 | 69.2 | 29.3 | 78.9 | 78.1 | 82.2 | 82.3 | 62.9 | |
| VLM2Vec#Params=4B, Latency=–2026.06 | 42.8 | 26.7 | 66.7 | 21.4 | 53.5 | 63.5 | 64 | 70.7 | 51.2 | |
| BM25#Params=–, Latency=–2026.06 | 40.1 | 38.4 | 70 | 61.5 | 88 | 84.7 | 82.7 | 89.2 | 69.3 | |
| BGE-M3#Params=569M, Latency=–2026.06 | 35.7 | 32.9 | 71.9 | 43.8 | 88.8 | 83.3 | 80.4 | 91.3 | 66 | |
| CLIP#Params=151M, Latency=–2026.06 | 26.5 | 14.6 | 51.7 | 4.7 | 22.9 | 32.4 | 39.8 | 37.5 | 28.8 | |
| Nomic-Embed-Vision#Params=92M, Latency=–2026.06 | 17.1 | 10.7 | 30.1 | 2.7 | 12.9 | 10.9 | 11.4 | 15.7 | 13.9 |