Hybrid search, multi-vector retrieval, custom ranking, and managed inference in one API. Built on object storage for 10x lower cost and massive scale.
1B+ docs
per collection
17ms p99
query latency
70MB/s
write throughput
Up to70%
higher recall
topk.collection("earnings_reports").query(select("content",# Semantic similaritysemantic = fn.semantic_similarity("content","NVDA data center revenue in Q4 2025"),# Multi-vector retrievalvisual = fn.multi_vector_distance("page_embedding",[[0.97, 0.17, ..], [0.14, 0.99, ..], ..]))# Keyword search.filter(match("nvidia") | match("nvda")),# Metadata filtering.filter(field("fiscal_year").eq(2025))# Custom scoring.topk((0.7 * field("semantic") + 0.3 * field("visual"))* field("source_quality"),10))
Inference
Unified Retrieval
TopK is a unified retrieval engine for search, RAG, and agents. Combine semantic, (multi) vector, and lexical search with metadata filters and custom scoring in a single query. No complex pipeline to build an manage.
TopK delivers state-of-the-art database performance and retrieval quality enabling more accurate answers while burning up to 10x fewer tokens.
Answer accuracy judged by GPT5 on Vidore V3 Finance
TopK comes with tooling to get you started fast. Python, Javascript, and Rust SDKs to integrate your app. CLI and MCP to connect your agents.
TopK is built from the ground up with enterprise security in mind. Data is encrypted in transit and at rest, access is scoped by role, and our infrastructure is audited continuously. When you need full control, we can deploy to your VPC or on-prem.

Start building for free. Move to production with usage-based pricing or private deployment.