OpenViSTAI · Agents · Blockchain · Crypto
Submit a project
Menu
RAG & Knowledge Systems

Casper

Casper is a Rust-built, self-hosted vector search database that stores and manages vectors and retrieves similar items through HNSW indexing. It is designed for high-load search systems and AI applications such as RAG, and exposes HTTP and gRPC APIs alongside Python, Go, and Rust clients.

Explore the practical guide ↓

Independent OpenViST project profile, based on linked sources.

Casper — OpenViST project profile in RAG & Knowledge Systems.
OpenViST project illustration. See the linked sources for product details.
Understand the project

Practical overview

When it fits

Use Casper when you need a self-hosted vector retrieval backend for RAG or similarity-search applications and want to assess HNSW indexing, quantization, and cluster options against your workload.

Who it is for

  • Developers building RAG systems that need vector storage and retrieval.
  • Teams creating high-load similarity-search or nearest-neighbor services.
  • Engineers evaluating a self-hosted vector database for image or recommendation retrieval.

Documented uses

  • Store and manage vectors used by an application’s retrieval pipeline.
  • Search collections for similar vectors using supported distance metrics.
  • Choose between f32 and i8 vector quantization for a workload’s precision and memory trade-offs.
  • Serve reads from replicas in the documented master/replica cluster topology.
Before you choose

Requirements & limits

  • The documented HNSW index supports Euclidean, L2SQ, cosine, and inner-product metrics.
  • F32 and i8 quantization offer different precision and memory/performance trade-offs.
  • HTTP and gRPC APIs cover collection management and vector operations; client libraries are listed for Python, Go, and Rust.
  • The documented cluster mode uses a write-accepting master and WAL-based replication to read-serving replicas.
  • The published Qdrant comparison uses a specified dataset, hardware, and index configuration; its throughput and recall figures should be read in that benchmark context.

Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.

Source-based practical guide

Start Casper from the Linux release archive

Download and launch the documented Casper release as a standalone server.

Before you start

  • A Linux x86_64 environment for the listed GNU release archive.
  • An API token set in the API_TOKEN environment variable.

Steps

  1. Download the v0.1.0 x86_64 GNU tar.gz release from the GitHub release URL.

  2. Extract the archive with tar.

  3. Set API_TOKEN to the token you will use.

  4. Run ./casper.

Expected outcome

The Casper server process starts using the downloaded executable.

Guide limitations

  • The repository’s quick-start steps do not document how to create collections or submit searches; consult its API documentation for those operations.
Documentation 1 ↗
COMPARE OPTIONS

Compare options for Vector search

These projects document a common task and share a project role. This is a source-based comparison, not an execution benchmark or a guarantee of interchangeability.

ProjectDocumented purposePlatformAccess and requirements
CasperStore and manage vectors used by an application’s retrieval pipeline.Self-hosted, Server
MettisIndex text documents through an HTTP endpoint.Linux, Self-hosted, Server
Evidence and transparency

Sources & review

Official repository ↗
Source record details
Source availability
Active
Sources checked
18 October 2026
Original website title
casper-vdb/Casper · Official homepage title · retained in its source language
How we review sourcesSuggest a correction

Related options

Qdrant — Named in Casper’s repository as the comparison system for its published throughput and recall benchmarks.

Explore project →