TopK

Embeddings + database + ranking in a single API.

Contact for Pricing

Description

TopK offers a comprehensive search solution delivered through a single API, integrating embeddings, a database, and sophisticated ranking capabilities. This approach aims to drastically simplify the development and maintenance of search functionalities by eliminating the need to integrate multiple providers or build complex data pipelines.

Designed with developer experience as a core principle, TopK facilitates seamless integration into existing tech stacks. It supports various search types including semantic, vector (dense/sparse), text (BM25), and hybrid retrieval, along with efficient filtering. The platform is built for scalability, handling billions of documents with high availability and offering both cloud and self-hosted deployment options for enhanced security and data control.

Key Features

Unified Search API: Combines embeddings, database, and ranking in one interface.
Hybrid Retrieval: Supports semantic, keyword, vector, and hybrid search.
Advanced Filtering: Efficient and flexible filtering on metadata and search parameters.
Reranking Engine: Built-in reranking to improve result relevance.
Vector Search: Handles dense and sparse vectors with quantization support.
Text Indexing: Implements BM25 scoring for text relevance.
Scalability: Designed to scale to billions of documents.
High Availability: Offers 99.9+% uptime across multiple regions.
Flexible Deployment: Available as a cloud service or for self-hosting.
Developer Focused: Prioritizes ease of integration and use.