Late Interaction Python Packages

topk-sdk

High-quality search for AI-native applications.

late-interaction-kernels

Fused Triton kernels for late-interaction (MaxSim) scoring

pylate-rs

PyLate efficient inference engine

latence-solver

High-performance late-interaction retrieval engine for on-prem AI. ColBERT/ColPali multi-vector search with Rust fused MaxSim, Triton GPU kernels, ROQ quantization, LEMUR routing, WAL-backed CRUD, and a FastAPI server — single machine, CPU or GPU.

raglite

🥤 RAGLite is a Python toolkit for Retrieval-Augmented Generation (RAG) with DuckDB or PostgreSQL

voyager-index

High-performance late-interaction retrieval engine for on-prem AI. ColBERT/ColPali multi-vector search with Rust fused MaxSim, Triton GPU kernels, ROQ quantization, LEMUR routing, WAL-backed CRUD, and a FastAPI server — single machine, CPU or GPU.

pymuvera

Python library for MUVERA multi-vector retrieval via Fixed Dimensional Encodings. ColBERT / ColQwen2 / ColQwen3.5 compatible.

pylate-rs-mkl

PyLate efficient inference engine

jina-v4-vllm-plugin

vLLM out-of-tree model: Jina Embeddings v4 multi-vector (token_embed) on Qwen2.5-VL

pylate-rs-metal

PyLate efficient inference engine

pylate-rs-accelerate

PyLate efficient inference engine

colsearch

High-performance late-interaction retrieval engine for on-prem AI. ColBERT/ColPali multi-vector search with Rust fused MaxSim, Triton GPU kernels, ROQ quantization, LEMUR routing, WAL-backed CRUD, and a FastAPI server — single machine, CPU or GPU.

pylate-rs-cuda

Rust and WebAssembly library for late interaction models.