Hanzo Vector Go SDK

Go client library for Hanzo Vector, speaking the Qdrant gRPC protocol. The main package is `qdrant` — types, client and helpers all live there.

Overview

Go client library for Hanzo Vector, speaking the Qdrant gRPC protocol. The main package is qdrant -- all types, client, and helpers live there.

Repo

hanzoai/vector-go, branch: master.

Note: the Go module path is still github.com/qdrant/go-client. That is what you put in your go.mod; github.com/hanzoai/vector-go is the repository, not the import path.

Quick reference

Item	Value
Module	`github.com/qdrant/go-client` (repo: `hanzoai/vector-go`)
Package	`qdrant`
Go	1.26+
Repo	`github.com/hanzoai/vector-go`
Branch	`master`
License	Apache-2.0
Protocol	gRPC
Default port	6334

Installation

go get github.com/hanzoai/vector-go

Quick start

package main

import (
    "context"
    "fmt"

    "github.com/hanzoai/vector-go/qdrant"
)

func main() {
    client, err := qdrant.NewClient(&qdrant.Config{
        Host: "localhost",
        Port: 6334,
    })
    if err != nil {
        panic(err)
    }
    defer client.Close()

    ctx := context.Background()

    // Create collection
    client.CreateCollection(ctx, &qdrant.CreateCollection{
        CollectionName: "docs",
        VectorsConfig: qdrant.NewVectorsConfig(&qdrant.VectorParams{
            Size:     384,
            Distance: qdrant.Distance_Cosine,
        }),
    })

    // Upsert points
    client.Upsert(ctx, &qdrant.UpsertPoints{
        CollectionName: "docs",
        Points: []*qdrant.PointStruct{
            {
                Id:      qdrant.NewIDNum(1),
                Vectors: qdrant.NewVectors(0.1, 0.2, 0.3),
                Payload: qdrant.NewValueMap(map[string]any{"title": "hello"}),
            },
        },
    })

    // Search
    results, _ := client.Query(ctx, &qdrant.QueryPoints{
        CollectionName: "docs",
        Query:          qdrant.NewQuery(0.1, 0.2, 0.3),
        WithPayload:    qdrant.NewWithPayload(true),
    })
    fmt.Println(results)
}

Key features

Collection management (create, update, delete, list)
Point upsert, delete, get, scroll, count
Vector search (query) with filtering
Payload filtering with match, range, geo conditions
Snapshot management (create, list, delete)
Connection pooling (configurable PoolSize, default 3, round-robin)
TLS and API key authentication
gRPC-based (protobuf generated types)

Client configuration

client, _ := qdrant.NewClient(&qdrant.Config{
    Host:   "xyz.cloud.qdrant.io",
    Port:   6334,
    APIKey: "<your-api-key>",
    UseTLS: true,
    // PoolSize: 3,          // gRPC connection pool size
    // TLSConfig: &tls.Config{},
    // GrpcOptions: []grpc.DialOption{},
})

Low-level gRPC access

// Direct gRPC service clients
collectionsClient := client.GetCollectionsClient()
pointsClient := client.GetPointsClient()
snapshotsClient := client.GetSnapshotsClient()
qdrantClient := client.GetQdrantClient()
conn := client.GetConnection() // raw *grpc.ClientConn

Helper functions

qdrant.NewIDNum(42)                    // numeric point ID
qdrant.NewIDUUID("uuid-string")       // UUID point ID
qdrant.NewVectors(0.1, 0.2, 0.3)     // dense vector
qdrant.NewVectorsConfig(params)       // collection vector config
qdrant.NewValueMap(map[string]any{})  // payload from map
qdrant.NewQuery(0.1, 0.2, 0.3)       // search query from floats
qdrant.NewWithPayload(true)           // include payload in results
qdrant.NewMatch("field", "value")     // filter condition
qdrant.PtrOf(value)                   // generic pointer helper

Build and test

go build ./...
go test ./...
# Integration tests use testcontainers (Docker required)

On this page