Visual document retrieval on ViDoRe V3
66.32HRnemotron-colembed-vl-8b-v2
Evaluation Results
| Method | Links | ||||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| nemotron-colembed-vl-8b-v2Rank=12026.02 | 66.32 | 67.29 | 56.03 | 67.19 | 79.3 | 69.82 | 50.84 | 51.54 | 63.42 | — | — | 53.84 | 72 | — | — | — | — | — | |
| tomoro-colqwen3-embed-8bRank=22026.02 | 63.98 | 65.08 | 54.41 | 66.36 | 75.35 | 68.41 | 50.13 | 49.1 | 61.59 | — | — | 52.65 | 70.46 | — | — | — | — | — | |
| nemotron-colembed-vl-4b-v2Rank=32026.02 | 62.39 | 65.02 | 53.91 | 66.1 | 78.56 | 67.48 | 48.86 | 49.01 | 61.54 | — | — | 52.78 | 71.3 | — | — | — | — | — | |
| llama-nemotron-colembed-vl-3b-v2Rank=62026.02 | 62.28 | 64.23 | 51.71 | 66.04 | 77.09 | 64.88 | 46.93 | 44.41 | 59.79 | — | — | 50.65 | 69.68 | — | — | — | — | — | |
| Ops-Colqwen3-4BRank=42026.02 | 61.81 | 65.71 | 53.99 | 66.42 | 77.74 | 66.49 | 49.14 | 48.81 | 61.17 | — | — | 52.23 | 69.33 | — | — | — | — | — | |
| tomoro-colqwen3-embed-4bRank=52026.02 | 60.09 | 63.84 | 53.58 | 65.74 | 75.44 | 66.43 | 49.32 | 46.83 | 60.2 | — | — | 51.23 | 69.44 | — | — | — | — | — | |
| jina-embeddings-v4Rank=72026.02 | 59.53 | 59.3 | 50.38 | 63.09 | 71.81 | 63.5 | 46.63 | 46.1 | 57.52 | — | — | 50.02 | 64.81 | — | — | — | — | — | |
| llama-nemoretriever-colembed-3b-v1Rank=92026.02 | 58.69 | 60.88 | 47.09 | 63.74 | 75.16 | 62.07 | 45.13 | 43.77 | 57.26 | — | — | 49.15 | 64.74 | — | — | — | — | — | |
| colnomic-embed-multimodal-7bRank=82026.02 | 58.67 | 56.57 | 50.13 | 62.26 | 76.2 | 63.58 | 48.25 | 45.46 | 57.33 | — | — | 45.02 | 67.16 | — | — | — | — | — | |
| BaseBase Architecture=ColQwen2.5, Token=49622026.01 | 47.3 | 50 | 41.6 | 56.1 | 68.6 | 57.1 | 43.4 | 37.5 | 50.2 | — | — | — | — | — | — | — | — | — | |
| CausalQwenBase Architecture=ColQwen2.5, Token=322026.01 | 42.9 | 39.5 | 32.7 | 48.3 | 64.3 | 46 | 39 | 28.5 | 42.6 | — | — | — | — | — | — | — | — | — | |
| K-MeansBase Architecture=ColQwen2.5, Token=322026.01 | 36.5 | 36 | 30.9 | 46.7 | 57.9 | 46.4 | 38.3 | 24.7 | 39.7 | — | — | — | — | — | — | — | — | — | |
| SemClusterBase Architecture=ColQwen2.5, Token=322026.01 | 35.9 | 36.6 | 30.7 | 48.1 | 59.4 | 45.3 | 39 | 26.7 | 40.2 | — | — | — | — | — | — | — | — | — | |
| CausalPaliBase Architecture=ColPali, Token=322026.01 | 35.3 | 28 | 23.4 | 43.6 | 53.2 | 35.3 | 33.2 | 16.9 | 33.6 | — | — | — | — | — | — | — | — | — | |
| BiQwen2Base Architecture=ColQwen2.5, Token=12026.01 | 33.6 | 33.3 | 23.6 | 46 | 55 | 39.9 | 38.3 | 20 | 36.2 | — | — | — | — | — | — | — | — | — | |
| BaseBase Architecture=ColPali, Token=10312026.01 | 32.1 | 23.2 | 22.8 | 39.1 | 48.8 | 24.8 | 24.8 | 14.4 | 28.8 | — | — | — | — | — | — | — | — | — | |
| 1D-PoolingBase Architecture=ColQwen2.5, Token=322026.01 | 26 | 28.3 | 23.7 | 40.4 | 56.1 | 39.7 | 37.9 | 18.6 | 33.8 | — | — | — | — | — | — | — | — | — | |
| SemClusterBase Architecture=ColPali, Token=322026.01 | 25.1 | 16.5 | 16.4 | 31.9 | 39 | 23.6 | 18.8 | 10.2 | 22.7 | — | — | — | — | — | — | — | — | — | |
| RandomBase Architecture=ColQwen2.5, Token=322026.01 | 24.8 | 25 | 20.3 | 38.4 | 49.2 | 34.8 | 33.8 | 14.4 | 30.1 | — | — | — | — | — | — | — | — | — | |
| K-MeansBase Architecture=ColPali, Token=322026.01 | 23 | 16.3 | 15.8 | 30.8 | 38 | 23.3 | 19.4 | 10 | 22.1 | — | — | — | — | — | — | — | — | — | |
| 1D-PoolingBase Architecture=ColPali, Token=322026.01 | 20.4 | 10.8 | 12.5 | 22.3 | 33.9 | 11.1 | 13.4 | 5.3 | 16.2 | — | — | — | — | — | — | — | — | — | |
| BiPaliBase Architecture=ColPali, Token=12026.01 | 19.7 | 10.4 | 10 | 18.2 | 37.2 | 15 | 17.5 | 7.1 | 16.9 | — | — | — | — | — | — | — | — | — | |
| RandomBase Architecture=ColPali, Token=322026.01 | 16.1 | 12.7 | 11.9 | 23.9 | 30.1 | 14.8 | 15.2 | 6 | 16.3 | — | — | — | — | — | — | — | — | — | |
| BGE-M3Size=0.6B, Retriever Type=Textual2026.03 | — | — | — | — | — | — | — | — | — | — | — | — | — | 58 | 52 | 28.5 | 46.2 | — | |
| BGE-M3 (caption)Size=0.57B, Late Interaction=true2026.05 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 38.8 | |
| BGE-M3 (unstr.)Size=0.57B, Late Interaction=true2026.05 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 45 | |
| BiModernVBERTSize=0.25B, Neural Query Encoder=true2026.05 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 28.9 | |
| BM25 (caption)FLOPs=1.32026.05 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 38.3 | |
| BM25 (unstr.)FLOPs=1.42026.05 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 38.7 | |
| ColEmbedSize=1B, Retriever Type=Visual2026.03 | — | — | — | — | — | — | — | — | — | — | — | — | — | 71.3 | 62.6 | 46.6 | 60.2 | — | |
| ColEmbedSize=3B, Retriever Type=Visual2026.03 | — | — | — | — | — | — | — | — | — | — | — | — | — | 75.2 | 63.7 | 47.1 | 62 | — | |
| ColEmbed-v2Size=3B, Retriever Type=Visual2026.03 | — | — | — | — | — | — | — | — | — | — | — | — | — | 77.1 | 66 | 51.7 | 64.9 | — | |
| ColFlorSize=0.17B, Late Interaction=true, Neural Query Encoder=true2026.05 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 36.8 | |
| ColModernVBERTSize=0.25B, Late Interaction=true, Neural Query Encoder=true2026.05 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 44.6 | |
| ColNomicSize=3B, Retriever Type=Visual2026.03 | — | — | — | — | — | — | — | — | — | — | — | — | — | 72.7 | 61.1 | 47.4 | 60.4 | — | |
| ColNomicSize=7B, Retriever Type=Visual2026.03 | — | — | — | — | — | — | — | — | — | — | — | — | — | 76.2 | 62.3 | 50.1 | 62.9 | — | |
| colnomic-embed-multimodal-3bSize=3B2026.01 | — | — | — | — | — | — | — | — | — | 56.4 | — | — | — | — | — | — | — | — | |
| colnomic-embed-multimodal-7bSize=7B2026.01 | — | — | — | — | — | — | — | — | — | 57.6 | — | — | — | — | — | — | — | — | |
| ColPaliSize=3B, Late Interaction=true, Neural Query Encoder=true2026.05 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 47.6 | |
| ColQwen2.5Size=3B, Retriever Type=Visual2026.03 | — | — | — | — | — | — | — | — | — | — | — | — | — | 72.3 | 57.9 | 41.3 | 57.2 | — | |
| colqwen2.5-v0.2Size=3B2026.01 | — | — | — | — | — | — | — | — | — | 52.4 | — | — | — | — | — | — | — | — | |
| D2SSize=0.25B, Neural Query Encoder=true, FLOPs=14.72026.05 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 1.8 | |
| E5-VSize=8B, Late Interaction=true2026.05 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 34.8 | |
| GME-Qwen2-7BSize=7B, Late Interaction=true2026.05 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 55.9 | |
| Jina CLIP v2Size=0.9B, Neural Query Encoder=true2026.05 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 25.7 | |
| jina-reranker-m0Size=2B2026.01 | — | — | — | — | — | — | — | — | — | — | 57.8 | — | — | — | — | — | — | — | |
| Jina-v4 (Textual)Size=3B, Retriever Type=Textual2026.03 | — | — | — | — | — | — | — | — | — | — | — | — | — | 64.3 | 54.9 | 38.4 | 52.5 | — | |
| Jina-v4 (Visual)Size=3B, Retriever Type=Visual2026.03 | — | — | — | — | — | — | — | — | — | — | — | — | — | 71.8 | 63.1 | 50.4 | 61.8 | — | |
| JSDO-SparseSize=0.25B, Neural Query Encoder=true, FLOPs=18142026.05 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 5.8 | |
| LITTASize=3B, Retriever Type=Visual, Query Expansion Count (q)=12026.03 | — | — | — | — | — | — | — | — | — | — | — | — | — | 79.3 | 74.3 | 39.05 | 64.2 | — | |
| LITTASize=3B, Retriever Type=Visual, Query Expansion Count (q)=32026.03 | — | — | — | — | — | — | — | — | — | — | — | — | — | 80.2 | 74.5 | 53.67 | 69.5 | — | |
| LITTASize=3B, Retriever Type=Visual, Query Expansion Count (q)=52026.03 | — | — | — | — | — | — | — | — | — | — | — | — | — | 80.2 | 74.3 | 61.39 | 72 | — | |
| llama-nemoretriever-colembed-1b-v1Size=1B2026.01 | — | — | — | — | — | — | — | — | — | 55.5 | — | — | — | — | — | — | — | — | |
| llama-nemoretriever-colembed-3b-v1Size=3B2026.01 | — | — | — | — | — | — | — | — | — | 57.1 | — | — | — | — | — | — | — | — | |
| NemColEmbed V2 8BSize=8B, Late Interaction=true, Neural Query Encoder=true2026.05 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 64.2 | |
| NomicSize=7B, Retriever Type=Visual2026.03 | — | — | — | — | — | — | — | — | — | — | — | — | — | 66.6 | 58.9 | 37.9 | 54.5 | — | |
| Qwen3Size=8B, Retriever Type=Textual2026.03 | — | — | — | — | — | — | — | — | — | — | — | — | — | 71.7 | 59.2 | 40.4 | 57.1 | — | |
| Qwen3-VL-Emb-2BSize=2B, Late Interaction=true2026.05 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 55 | |
| Qwen3-VL-Embedding-2BSize=2B2026.01 | — | — | — | — | — | — | — | — | — | 52.9 | — | — | — | — | — | — | — | — | |
| Qwen3-VL-Embedding-2BSize=2B2026.01 | — | — | — | — | — | — | — | — | — | — | 52.9 | — | — | — | — | — | — | — | |
| Qwen3-VL-Embedding-8BSize=8B2026.01 | — | — | — | — | — | — | — | — | — | 59 | — | — | — | — | — | — | — | — | |
| Qwen3-VL-Ranker-2BSize=2B2026.01 | — | — | — | — | — | — | — | — | — | 60.8 | — | — | — | — | — | — | — | — | |
| Qwen3-VL-Ranker-8BSize=8B2026.01 | — | — | — | — | — | — | — | — | — | 66.7 | — | — | — | — | — | — | — | — | |
| Qwen3-VL-Reranker-2BSize=2B2026.01 | — | — | — | — | — | — | — | — | — | — | 60.8 | — | — | — | — | — | — | — | |
| Qwen3-VL-Reranker-8BSize=8B2026.01 | — | — | — | — | — | — | — | — | — | — | 66.7 | — | — | — | — | — | — | — | |
| Qwen3.5Size=8B, Retriever Type=Textual2026.03 | — | — | — | — | — | — | — | — | — | — | — | — | — | 73.7 | 67 | 42.6 | 61.1 | — | |
| SigLIP2-LSize=0.9B, Neural Query Encoder=true2026.05 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 22.9 | |
| tomoro-colqwen3-embed-4bSize=4B2026.01 | — | — | — | — | — | — | — | — | — | 60.2 | — | — | — | — | — | — | — | — | |
| tomoro-colqwen3-embed-8bSize=8B2026.01 | — | — | — | — | — | — | — | — | — | 61.6 | — | — | — | — | — | — | — | — | |
| V-SPLADE + BiModernVBERTSize=0.25/0.25B, Neural Query Encoder=true2026.05 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 43.6 | |
| V-SPLADE + BiModernVBERT + BM25 (unstr.)Size=0.25/0.25B, Neural Query Encoder=true2026.05 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 47.2 | |
| V-SPLADE + BM25 (unstr.)Size=0.25B2026.05 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 44.8 | |
| V-SPLADE efficientSize=0.25B, FLOPs=1.12026.05 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 37.6 | |
| V-SPLADE qualitySize=0.25B, FLOPs=1.92026.05 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 40.9 | |
| VLM2VecSize=4B, Late Interaction=true2026.05 | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | — | 26.3 |