ColServe
Late-interaction visual document retrieval in 5 lines of Python.
from colserve import Index, Retriever
idx = Index(model="colsmol-256m", storage="qdrant://localhost:6333")
idx.add_pdf("report.pdf")
retriever = Retriever(idx)
results = retriever.search("revenue by region", top_k=3)Open-source Python library and hosted service for indexing PDFs, scanned documents, and other visually-rich content. Powered by the Col-family of multi-vector retrieval models (ColPali, ColQwen2, ColSmol).
Launching in 8 weeks. Get notified when it ships.
After signing up, check your inbox and click the confirmation link. No spam — one email when ColServe is ready.