TomoroAI/tomoro-colqwen3-embed-4b
Primitive: /encode · Encode ·
Qwen3-VL
TomoroAI/tomoro-colqwen3-embed-4b is a state-of-the-art ColPali-style multimodal embedding model. It maps text queries, visual documents (images, PDFs) or short videos into aligned multi-vector embeddings.
MultimodalMultilingualLong contextMulti-vector
View on Hugging Face → Fine-tuned from Qwen/Qwen3-VL-4B-Instruct
Overview
Hardware: — drives latency, throughput & cost
| Size | 4.4B params |
|---|---|
| Tasks | /encode |
| License | apache-2.0 |
| Languages | multilingual |
| Latency | — |
| Throughput | — |
| Cost | — /1M tok |
Cost is approximate — computed from list GPU prices; your actual price depends on the provider you deploy SIE with.
Embedding
| Output types | Multi-Vec |
|---|---|
| Dimensions | multivector: 320 |
| Max sequence length | 8,192 |
| Inputs | text · image |
Benchmarks
Vidore3ComputerScienceRetrieval
Visual document retrieval on computer science papers and slides
default_lang-eng
Quality
map at 10 0.6491
mrr at 10 0.8847
ndcg at 10 0.7690
muvera_lang-eng_limit-256
Quality
ndcg at 10 0.8558
map at 10 0.7873
mrr at 10 0.9134
Vidore3FinanceEnRetrieval
Visual document retrieval on financial reports
default_lang-eng
Quality
map at 10 0.5816
mrr at 10 0.8172
ndcg at 10 0.6970
muvera_lang-eng_limit-256
Quality
ndcg at 10 0.7161
map at 10 0.6000
mrr at 10 0.7717
Vidore3HrRetrieval
Visual document retrieval on HR-related documents
Quality
ndcg at 10 0.5851
map at 10 0.4758
mrr at 10 0.6364
Vidore3PharmaceuticalsRetrieval
Visual document retrieval on pharmaceutical documents
default_lang-eng
Quality
map at 10 0.5754
mrr at 10 0.7978
ndcg at 10 0.6852
muvera_lang-eng_limit-256
Quality
ndcg at 10 0.7916
map at 10 0.6890
mrr at 10 0.8737
Compare (0)Compare →