### Install Qdrant PHP Client

Source: https://github.com/hkulekci/qdrant-php/blob/main/README.md

Install the Qdrant PHP client using composer. This is the first step to integrating Qdrant into your PHP project.

```shell
composer require hkulekci/qdrant
```

--------------------------------

### Install Qdrant PHP Client with Composer

Source: https://context7.com/hkulekci/qdrant-php/llms.txt

Install the Qdrant PHP client and required dependencies using Composer. Ensure you have an HTTP client like Symfony HTTP Client and a PSR-7 factory like Nyholm PSR7.

```bash
composer require hkulekci/qdrant symfony/http-client nyholm/psr7
```

--------------------------------

### Create a Qdrant Collection

Source: https://github.com/hkulekci/qdrant-php/blob/main/README.md

Create a new collection in Qdrant with specified vector parameters. This example defines a collection named 'contents' with cosine distance for 1536-dimensional vectors.

```php
use Qdrant\Endpoints\Collections;
use Qdrant\Models\Request\CreateCollection;
use Qdrant\Models\Request\VectorParams;

$createCollection = new CreateCollection();
$createCollection->addVector(new VectorParams(1536, VectorParams::DISTANCE_COSINE), 'content');
$response = $client->collections('contents')->create($createCollection);
```

--------------------------------

### Query Points with `Query::query()` - Recommendation

Source: https://context7.com/hkulekci/qdrant-php/llms.txt

Execute recommendation queries using positive and negative example point IDs. This method replaces the deprecated `recommend()` method and requires importing `QueryRequest`.

```php
use Qdrant\Models\Request\Points\QueryRequest;

// --- Recommendation query (positive/negative examples) ---
$recRequest = (new QueryRequest())
    ->setQuery([
        'recommend' => [
            'positive' => [1, 5],   // point IDs to use as positive examples
            'negative' => [3],      // point IDs to push away from
        ]
    ])
    ->setUsing('content')
    ->setLimit(5)
    ->setWithPayload(true);

$recResponse = $client->collections('articles')->points()->query()->query($recRequest);
```

--------------------------------

### List, Inspect, and Check Collection Existence

Source: https://context7.com/hkulekci/qdrant-php/llms.txt

Provides methods to list all collections, get detailed information about a specific collection, and check if a collection exists.

```APIDOC
## List and Inspect Collections — `Collections::list()`, `Collections::info()`, `Collections::exists()`

Retrieve all collections, inspect a single collection's configuration and statistics, or check existence.

```php
// List all collections
$all = $client->collections()->list();
foreach ($all['result']['collections'] as $col) {
    echo $col['name'] . PHP_EOL;
}

// Get detailed info for a specific collection
$info = $client->collections('articles')->info();
echo $info['result']['status'];         // 'green'
echo $info['result']['vectors_count'];  // total vectors stored

// Check existence
$exists = $client->collections('articles')->exists();
// $exists['result']['exists'] === true|false
```
```

--------------------------------

### Search Points with OpenAI Embeddings

Source: https://github.com/hkulekci/qdrant-php/blob/main/README.md

Search for points using embeddings generated by OpenAI. This example demonstrates generating embeddings for a query and then performing a search, followed by iterating through results.

```php
$openai = OpenAI::client(OPENAI_API_KEY);

$query = 'lorem ipsum dolor sit amed';
$response = $openai->embeddings()->create([
    'model' => 'text-embedding-ada-002',
    'input' => $query,
]);
$embedding = array_values($response->embeddings[0]->embedding);

$searchRequest = (new SearchRequest(new VectorStruct($embedding, 'content')))
    ->setLimit(10)
    ->setParams([
        'hnsw_ef' => 128,
        'exact' => false,
    ])
    ->setWithPayload(true);

$response = $client->collections('contents')->points()->search($searchRequest);

foreach ($response['result'] as $item) {
    echo $item['score'] . ';' . $item['payload']['id'] . ';' . $item['payload']['meta_data'] . PHP_EOL;
}
```

--------------------------------

### List, Inspect, and Check Collection Existence

Source: https://context7.com/hkulekci/qdrant-php/llms.txt

Retrieve a list of all collections, get detailed information about a specific collection, or check if a collection exists. Useful for managing and monitoring collections.

```php
// List all collections
$all = $client->collections()->list();
foreach ($all['result']['collections'] as $col) {
    echo $col['name'] . PHP_EOL;
}

// Get detailed info for a specific collection
$info = $client->collections('articles')->info();
echo $info['result']['status'];         // 'green'
echo $info['result']['vectors_count'];  // total vectors stored

// Check existence
$exists = $client->collections('articles')->exists();
// $exists['result']['exists'] === true|false
```

--------------------------------

### Search Points with Filter

Source: https://github.com/hkulekci/qdrant-php/blob/main/README.md

Perform a search operation on points within a collection, applying a filter to narrow down results. This example searches for points matching 'Palm' in the 'name' field.

```php
use Qdrant\Models\Filter\Condition\MatchString;
use Qdrant\Models\Filter\Filter;
use Qdrant\Models\Request\SearchRequest;
use Qdrant\Models\VectorStruct;

$searchRequest = (new SearchRequest(new VectorStruct($embedding, 'elev_pitch')))
    ->setFilter(
        (new Filter())->addMust(
            new MatchString('name', 'Palm')
        )
    )
    ->setLimit(10)
    ->setParams([
        'hnsw_ef' => 128,
        'exact' => false,
    ])
    ->setWithPayload(true);

$response = $client->collections('contents')->points()->search($searchRequest);
```

--------------------------------

### Count Points

Source: https://context7.com/hkulekci/qdrant-php/llms.txt

Get the total number of points in a collection or count points that match a specified filter. This is useful for understanding collection size or subset sizes.

```APIDOC
## Count Points — `Points::count()`

Count total points or points matching a filter.

```php
use Qdrant\Models\Filter\Filter;
use Qdrant\Models\Filter\Condition\MatchString;

// Count all
$response = $client->collections('articles')->points()->count();
echo $response['result']['count']; // e.g. 10540

// Count with filter (exact count)
$filter = (new Filter())->addMust(new MatchString('tags', 'eco'));
$response = $client->collections('articles')->points()->count($filter, exact: true);
echo $response['result']['count'];
```
```

--------------------------------

### Count Points — `Points::count()`

Source: https://context7.com/hkulekci/qdrant-php/llms.txt

Get the total number of points in a collection or the count of points that match a given filter. The `exact` parameter can be set to `true` for a precise count, which may be slower for large collections.

```php
use Qdrant\Models\Filter\Filter;
use Qdrant\Models\Filter\Condition\MatchString;

// Count all
$response = $client->collections('articles')->points()->count();
echo $response['result']['count']; // e.g. 10540

// Count with filter (exact count)
$filter = (new Filter())->addMust(new MatchString('tags', 'eco'));
$response = $client->collections('articles')->points()->count($filter, exact: true);
echo $response['result']['count'];
```

--------------------------------

### Connect to Qdrant

Source: https://github.com/hkulekci/qdrant-php/blob/main/README.md

Establish a connection to the Qdrant instance using configuration details like host and API key. Ensure you have the `config.php` file with necessary credentials.

```php
include __DIR__ . "/../vendor/autoload.php";
include_once 'config.php';

use Qdrant\Qdrant;
use Qdrant\Config;
use Qdrant\Http\Builder;

$config = new Config(QDRANT_HOST);
$config->setApiKey(QDRANT_API_KEY);

$transport = (new Builder())->build($config);
$client = new Qdrant($transport);
```

--------------------------------

### Client Initialization

Source: https://context7.com/hkulekci/qdrant-php/llms.txt

Initializes the Qdrant client with connection parameters and an optional API key.

```APIDOC
## Client Initialization — `Qdrant`, `Config`, `Http\Builder`

`Config` holds the connection parameters (host, port, optional API key). `Http\Builder` wires together the PSR-18 HTTP client and the config into a `Transport`, which is passed to the main `Qdrant` client.

```php
use Qdrant\Qdrant;
use Qdrant\Config;
use Qdrant\Http\Builder;

$config = new Config('http://localhost', 6333);
$config->setApiKey('your-api-key'); // optional, for Qdrant Cloud

$transport = (new Builder())->build($config);
$client    = new Qdrant($transport);
```
```

--------------------------------

### Initialize Qdrant Client

Source: https://context7.com/hkulekci/qdrant-php/llms.txt

Initialize the Qdrant client by configuring connection parameters and building the transport layer. The API key is optional and used for Qdrant Cloud.

```php
use Qdrant\Qdrant;
use Qdrant\Config;
use Qdrant\Http\Builder;

$config = new Config('http://localhost', 6333);
$config->setApiKey('your-api-key'); // optional, for Qdrant Cloud

$transport = (new Builder())->build($config);
$client    = new Qdrant($transport);
```

--------------------------------

### Create and Manage Custom Sharding Keys

Source: https://context7.com/hkulekci/qdrant-php/llms.txt

Use these methods to create and delete custom sharding keys for tenant-based data isolation. Ensure the `CreateShardKey` or `DeleteShardKey` models are imported.

```php
use Qdrant\Models\Request\CollectionConfig\CreateShardKey;
use Qdrant\Models\Request\CollectionConfig\DeleteShardKey;

// Create a custom shard key
$createShard = new CreateShardKey(
    shardKey:          'tenant-001',
    shardNumber:        2,
    replicationFactor:  1,
    placement:          [1, 2]   // peer IDs
);
$client->collections('articles')->shards()->create($createShard);

// Delete a shard key
$deleteShard = new DeleteShardKey('tenant-001');
$client->collections('articles')->shards()->delete($deleteShard);
```

--------------------------------

### Manage Collection Snapshots

Source: https://context7.com/hkulekci/qdrant-php/llms.txt

Use these methods to create, list, download, restore, and delete snapshots for individual collections. Ensure the 'wait' parameter is set to 'true' for synchronous operations.

```php
// Create a snapshot
$response = $client->collections('articles')->snapshots()->create([], ['wait' => 'true']);
$snapshotName = $response['result']['name']; // e.g. "articles-1715000000"
```

```php
// List snapshots
$list = $client->collections('articles')->snapshots()->list();
foreach ($list['result'] as $snap) {
    echo $snap['name'] . ' size=' . $snap['size'] . PHP_EOL;
}
```

```php
// Download snapshot (returns binary stream response)
$download = $client->collections('articles')->snapshots()->download($snapshotName);
```

```php
// Restore (recover) a collection from snapshot
$recover = $client->collections('articles')->snapshots()->recover();
```

```php
// Delete snapshot
$client->collections('articles')->snapshots()->delete($snapshotName);
```

--------------------------------

### Manage Collection Aliases

Source: https://context7.com/hkulekci/qdrant-php/llms.txt

Manage zero-downtime collection aliases for blue/green deployments. Create, switch, or delete aliases atomically.

```php
use Qdrant\Models\Request\AliasActions;

// Create or switch an alias
$actions = new AliasActions();
$actions->add('articles-live', 'articles-v2');  // alias → collection
$actions->delete('articles-live');              // remove old alias (atomic swap)

$response = $client->collections('articles-v2')->aliases()->actions($actions);
```

```php
// List all aliases on a collection
$aliases = $client->collections('articles-v2')->aliases()->aliases();
```

```php
// List all aliases globally
$allAliases = $client->aliases()->all();
```

--------------------------------

### Manage Storage-Level Snapshots

Source: https://context7.com/hkulekci/qdrant-php/llms.txt

These methods manage full-storage snapshots that encompass all collections. Use 'wait' => 'true' for synchronous snapshot creation.

```php
// List storage snapshots
$list = $client->snapshots()->get();
```

```php
// Create a storage-wide snapshot
$response = $client->snapshots()->create(['wait' => 'true']);
$name = $response['result']['name'];
```

```php
// Download storage snapshot
$client->snapshots()->download($name);
```

```php
// Delete storage snapshot
$client->snapshots()->delete($name);
```

--------------------------------

### Storage-Level Snapshots

Source: https://context7.com/hkulekci/qdrant-php/llms.txt

Manage full-storage snapshots that cover all collections within the Qdrant instance.

```APIDOC
## Storage-Level Snapshots — `Snapshots`

### Description
Manage full-storage snapshots covering all collections.

### List Storage Snapshots
```php
$list = $client->snapshots()->get();
```

### Create Storage Snapshot
```php
$response = $client->snapshots()->create(['wait' => 'true']);
$name = $response['result']['name'];
```

### Download Storage Snapshot
```php
$client->snapshots()->download($name);
```

### Delete Storage Snapshot
```php
$client->snapshots()->delete($name);
```
```

--------------------------------

### Create a New Collection

Source: https://context7.com/hkulekci/qdrant-php/llms.txt

Define and create a new Qdrant collection with specified vector parameters, distance metric, HNSW configuration, and quantization settings. Supports named vector spaces.

```php
use Qdrant\Models\Request\CreateCollection;
use Qdrant\Models\Request\VectorParams;
use Qdrant\Models\Request\CollectionConfig\HnswConfig;
use Qdrant\Models\Request\CollectionConfig\ScalarQuantization;

$createCollection = new CreateCollection();
$createCollection->addVector(
    new VectorParams(1536, VectorParams::DISTANCE_COSINE),
    'content'                        // named vector space
);
$createCollection->addVector(
    new VectorParams(512, VectorParams::DISTANCE_EUCLID),
    'thumbnail'
);
$createCollection->setHnswConfig(
    (new HnswConfig())->setM(16)->setEfConstruct(100)
);
$createCollection->setQuantizationConfig(
    new ScalarQuantization('int8', 0.99, true)  // type, quantile, always_ram
);
$createCollection->setReplicationFactor(2);
$createCollection->setOnDiskPayload(false);

$response = $client->collections('articles')->create($createCollection);
// $response['status'] === 'ok'
// $response['result'] === true
```

--------------------------------

### Inspect Cluster Topology and Manage Shards

Source: https://context7.com/hkulekci/qdrant-php/llms.txt

Provides methods for inspecting cluster topology and performing shard operations like moving, replicating, or dropping shards. Ensure necessary models are imported.

```php
use Qdrant\Models\Request\UpdateCollectionCluster;
use Qdrant\Models\Request\ClusterUpdate\MoveShardOperation;
use Qdrant\Models\Request\ClusterUpdate\ReplicateShardOperation;
use Qdrant\Models\Request\ClusterUpdate\DropReplicaOperation;

// Global cluster status
$clusterInfo = $client->cluster()->info();

// Collection-level cluster info
$colCluster = $client->collections('articles')->cluster()->info();

// Move shard between peers
$moveOp = new MoveShardOperation(
    shardId:    0,
    toPeerId:   2,
    fromPeerId: 1
);
$client->collections('articles')->cluster()->update(
    new UpdateCollectionCluster($moveOp)
);

// Replicate a shard to another peer
$replicateOp = new ReplicateShardOperation(
    shardId:    0,
    toPeerId:   3,
    fromPeerId: 1
);
$client->collections('articles')->cluster()->update(
    new UpdateCollectionCluster($replicateOp)
);

// Drop a replica
$dropOp = new DropReplicaOperation(shardId: 0, peerId: 3);
$client->collections('articles')->cluster()->update(
    new UpdateCollectionCluster($dropOp)
);

// Remove a peer from the cluster
$client->cluster()->removePeer(peerId: 2, queryParams: ['force' => 'true']);
```

--------------------------------

### Collection Aliases

Source: https://context7.com/hkulekci/qdrant-php/llms.txt

Manage zero-downtime collection aliases for blue/green deployments using `Aliases::actions()` and `Aliases::aliases()`.

```APIDOC
## Collection Aliases — `Aliases::actions()`, `Aliases::aliases()`

Manage zero-downtime collection aliases for blue/green deployments.

```php
use Qdrant\Models\Request\AliasActions;

// Create or switch an alias
$actions = new AliasActions();
$actions->add('articles-live', 'articles-v2');  // alias → collection
$actions->delete('articles-live');              // remove old alias (atomic swap)

$response = $client->collections('articles-v2')->aliases()->actions($actions);

// List all aliases on a collection
$aliases = $client->collections('articles-v2')->aliases()->aliases();

// List all aliases globally
$allAliases = $client->aliases()->all();
```
```

--------------------------------

### Collection Snapshots

Source: https://context7.com/hkulekci/qdrant-php/llms.txt

Manage snapshots for individual collections, including creation, listing, downloading, restoring, and deletion.

```APIDOC
## Collection Snapshots — `Collections\Snapshots`

### Description
Create, list, download, and restore collection snapshots for backup and migration.

### Create Snapshot
```php
$response = $client->collections('articles')->snapshots()->create([], ['wait' => 'true']);
$snapshotName = $response['result']['name'];
```

### List Snapshots
```php
$list = $client->collections('articles')->snapshots()->list();
foreach ($list['result'] as $snap) {
    echo $snap['name'] . ' size=' . $snap['size'] . PHP_EOL;
}
```

### Download Snapshot
```php
$download = $client->collections('articles')->snapshots()->download($snapshotName);
```

### Restore Snapshot
```php
$recover = $client->collections('articles')->snapshots()->recover();
```

### Delete Snapshot
```php
$client->collections('articles')->snapshots()->delete($snapshotName);
```
```

--------------------------------

### Batch Upsert Points with `Points::batch()`

Source: https://context7.com/hkulekci/qdrant-php/llms.txt

Efficiently insert multiple points using the columnar batch format for better performance. Set `wait` to 'true' for synchronous acknowledgment. Requires importing `PointsBatch`, `PointStruct`, and `VectorStruct`.

```php
use Qdrant\Models\Request\PointsBatch;
use Qdrant\Models\PointStruct;
use Qdrant\Models\VectorStruct;

$batch = new PointsBatch();
$batch->addPoint(PointStruct::createFromArray([
    'id'      => 1,
    'vector'  => new VectorStruct([0.1, 0.2, 0.3], 'image'),
    'payload' => ['color' => 'red', 'size' => 'large'],
]));
$batch->addPoint(PointStruct::createFromArray([
    'id'      => 2,
    'vector'  => new VectorStruct([0.4, 0.5, 0.6], 'image'),
    'payload' => ['color' => 'blue', 'size' => 'small'],
]));
$batch->addPoint(PointStruct::createFromArray([
    'id'      => 3,
    'vector'  => new VectorStruct([0.7, 0.8, 0.9], 'image'),
    'payload' => ['color' => 'green', 'size' => 'medium'],
]));

$response = $client->collections('products')->points()->batch($batch, ['wait' => 'true']);
// $response['status'] === 'ok'
```

--------------------------------

### Cluster Info and Management

Source: https://context7.com/hkulekci/qdrant-php/llms.txt

Inspect cluster topology and perform shard operations across the Qdrant cluster.

```APIDOC
## Cluster Info and Management — `Cluster`, `Collections\Cluster`

### Description
Inspect cluster topology and perform shard operations.

### Global Cluster Info
```php
$clusterInfo = $client->cluster()->info();
```

### Collection-Level Cluster Info
```php
$colCluster = $client->collections('articles')->cluster()->info();
```

### Move Shard
```php
use Qdrant\Models\Request\UpdateCollectionCluster;
use Qdrant\Models\Request\ClusterUpdate\MoveShardOperation;

$moveOp = new MoveShardOperation(
    shardId:    0,
    toPeerId:   2,
    fromPeerId: 1
);
$client->collections('articles')->cluster()->update(
    new UpdateCollectionCluster($moveOp)
);
```

### Replicate Shard
```php
use Qdrant\Models\Request\ClusterUpdate\ReplicateShardOperation;

$replicateOp = new ReplicateShardOperation(
    shardId:    0,
    toPeerId:   3,
    fromPeerId: 1
);
$client->collections('articles')->cluster()->update(
    new UpdateCollectionCluster($replicateOp)
);
```

### Drop Replica
```php
use Qdrant\Models\Request\ClusterUpdate\DropReplicaOperation;

$dropOp = new DropReplicaOperation(shardId: 0, peerId: 3);
$client->collections('articles')->cluster()->update(
    new UpdateCollectionCluster($dropOp)
);
```

### Remove Peer
```php
$client->cluster()->removePeer(peerId: 2, queryParams: ['force' => 'true']);
```
```

--------------------------------

### Query Points

Source: https://context7.com/hkulekci/qdrant-php/llms.txt

The universal query endpoint supporting nearest-neighbor search, recommendation by point IDs, discover, fusion, sample, and multi-stage prefetch. Replaces the deprecated `search()` and `recommend()` methods.

```APIDOC
## Query Points (Recommended) — `Query::query()`

The universal query endpoint supporting nearest-neighbor search, recommendation by point IDs, discover, fusion, sample, and multi-stage prefetch. Replaces the deprecated `search()` and `recommend()` methods.

```php
use Qdrant\Models\Request\Points\QueryRequest;
use Qdrant\Models\Filter\Filter;
use Qdrant\Models\Filter\Condition\MatchString;
use Qdrant\Models\Filter\Condition\Range;

// --- Nearest-neighbor vector search ---
$queryRequest = (new QueryRequest())
    ->setQuery(['nearest' => [0.12, 0.53, /* ...1536 dims */]])
    ->setUsing('content')                          // named vector space
    ->setFilter(
        (new Filter())
            ->addMust(new MatchString('tags', 'eco'))
            ->addMust(new Range('year', ['gte' => 2020]))
    )
    ->setParams(['hnsw_ef' => 128, 'exact' => false])
    ->setScoreThreshold(0.7)
    ->setLimit(10)
    ->setOffset(0)
    ->setWithPayload(true)
    ->setWithVector(false);

$response = $client->collections('articles')->points()->query()->query($queryRequest);

foreach ($response['result']['points'] as $point) {
    echo $point['id'] . ' score=' . $point['score'] . ' title=' . $point['payload']['title'] . PHP_EOL;
}

// --- Recommendation query (positive/negative examples) ---
$recRequest = (new QueryRequest())
    ->setQuery([
        'recommend' => [
            'positive' => [1, 5],   // point IDs to use as positive examples
            'negative' => [3],      // point IDs to push away from
        ]
    ])
    ->setUsing('content')
    ->setLimit(5)
    ->setWithPayload(true);

$recResponse = $client->collections('articles')->points()->query()->query($recRequest);
```
```

--------------------------------

### Insert Points into Collection

Source: https://github.com/hkulekci/qdrant-php/blob/main/README.md

Insert points into a Qdrant collection. This snippet includes generating embeddings using OpenAI and structuring the data for insertion.

```php
use Qdrant\Models\PointsStruct;
use Qdrant\Models\PointStruct;
use Qdrant\Models\VectorStruct;

$openai = OpenAI::client(OPENAI_API_KEY);

$query = 'sustainable agricultural startups';
$response = $openai->embeddings()->create([
    'model' => 'text-embedding-ada-002',
    'input' => $query,
]);
$embedding = array_values($response->embeddings[0]->embedding);

$points = new PointsStruct();
$points->addPoint(
    new PointStruct(
        (int) $imageId,
        new VectorStruct($embedding, 'content'),
        [
            'id' => 1,
            'meta' => 'Meta data'
        ]
    )
);
$client->collections('contents')->points()->upsert($points);
```

--------------------------------

### Shards (Custom Sharding)

Source: https://context7.com/hkulekci/qdrant-php/llms.txt

Create and manage custom sharding keys for tenant-based data isolation within collections.

```APIDOC
## Shards (Custom Sharding) — `Shards::create()`, `Shards::delete()`

### Description
Create and manage custom sharding keys for tenant-based data isolation.

### Create Shard Key
```php
use Qdrant\Models\Request\CollectionConfig\CreateShardKey;

$createShard = new CreateShardKey(
    shardKey:          'tenant-001',
    shardNumber:        2,
    replicationFactor:  1,
    placement:          [1, 2]   // peer IDs
);
$client->collections('articles')->shards()->create($createShard);
```

### Delete Shard Key
```php
use Qdrant\Models\Request\CollectionConfig\DeleteShardKey;

$deleteShard = new DeleteShardKey('tenant-001');
$client->collections('articles')->shards()->delete($deleteShard);
```
```

--------------------------------

### Create Collection

Source: https://context7.com/hkulekci/qdrant-php/llms.txt

Defines and creates a new named vector collection with specified parameters.

```APIDOC
## Create a Collection — `Collections::create()`

Defines a new named vector collection with one or more named vector spaces, distance metric, optional quantization, HNSW tuning, and WAL configuration.

```php
use Qdrant\Models\Request\CreateCollection;
use Qdrant\Models\Request\VectorParams;
use Qdrant\Models\Request\CollectionConfig\HnswConfig;
use Qdrant\Models\Request\CollectionConfig\ScalarQuantization;

$createCollection = new CreateCollection();
$createCollection->addVector(
    new VectorParams(1536, VectorParams::DISTANCE_COSINE),
    'content'                        // named vector space
);
$createCollection->addVector(
    new VectorParams(512, VectorParams::DISTANCE_EUCLID),
    'thumbnail'
);
$createCollection->setHnswConfig(
    (new HnswConfig())->setM(16)->setEfConstruct(100)
);
$createCollection->setQuantizationConfig(
    new ScalarQuantization('int8', 0.99, true)  // type, quantile, always_ram
);
$createCollection->setReplicationFactor(2);
$createCollection->setOnDiskPayload(false);

$response = $client->collections('articles')->create($createCollection);
// $response['status'] === 'ok'
// $response['result'] === true
```
```

--------------------------------

### Query Points with `Query::query()` - Nearest Neighbor Search

Source: https://context7.com/hkulekci/qdrant-php/llms.txt

Perform nearest-neighbor vector search using the universal query endpoint. Supports filtering, named vector spaces, and score thresholds. Requires importing `QueryRequest`, `Filter`, `MatchString`, and `Range`.

```php
use Qdrant\Models\Request\Points\QueryRequest;
use Qdrant\Models\Filter\Filter;
use Qdrant\Models\Filter\Condition\MatchString;
use Qdrant\Models\Filter\Condition\Range;

// --- Nearest-neighbor vector search ---
$queryRequest = (new QueryRequest())
    ->setQuery(['nearest' => [0.12, 0.53, /* ...1536 dims */]])
    ->setUsing('content')                          // named vector space
    ->setFilter(
        (new Filter())
            ->addMust(new MatchString('tags', 'eco'))
            ->addMust(new Range('year', ['gte' => 2020]))
    )
    ->setParams(['hnsw_ef' => 128, 'exact' => false])
    ->setScoreThreshold(0.7)
    ->setLimit(10)
    ->setOffset(0)
    ->setWithPayload(true)
    ->setWithVector(false);

$response = $client->collections('articles')->points()->query()->query($queryRequest);

foreach ($response['result']['points'] as $point) {
    echo $point['id'] . ' score=' . $point['score'] . ' title=' . $point['payload']['title'] . PHP_EOL;
}
```

--------------------------------

### Wait for Upsert Acknowledgement

Source: https://github.com/hkulekci/qdrant-php/blob/main/README.md

Ensure data upsert operations are completed by Qdrant before proceeding. This is achieved by passing the 'wait' query parameter set to 'true'.

```php
$client->collections('contents')->points()->upsert($points, ['wait' => 'true']);
```

--------------------------------

### Service Telemetry and Metrics

Source: https://context7.com/hkulekci/qdrant-php/llms.txt

Retrieve Prometheus-compatible metrics or anonymized telemetry from the Qdrant instance.

```APIDOC
## Service Telemetry and Metrics — `Service`

### Description
Retrieve Prometheus-compatible metrics or anonymized telemetry from the Qdrant instance.

### Collect Telemetry
```php
$telemetry = $client->service()->telemetry(anonymize: true);
```

### Collect Metrics
```php
$metrics = $client->service()->metrics(anonymize: false);
// Returns Prometheus text exposition format
```
```

--------------------------------

### Retrieve Service Telemetry and Metrics

Source: https://context7.com/hkulekci/qdrant-php/llms.txt

Collect anonymized telemetry or Prometheus-compatible metrics from the Qdrant instance. The `metrics` method returns data in Prometheus text exposition format.

```php
// Collect anonymized telemetry
$telemetry = $client->service()->telemetry(anonymize: true);

// Collect Prometheus metrics
$metrics = $client->service()->metrics(anonymize: false);
// Returns Prometheus text exposition format
```

--------------------------------

### Multi-Vector Points with `MultiVectorStruct`

Source: https://context7.com/hkulekci/qdrant-php/llms.txt

Insert points into collections that support multiple named vector spaces. Requires importing `MultiVectorStruct`, `PointStruct`, and `PointsStruct`. The `upsert` method is used after defining the `MultiVectorStruct`.

```php
use Qdrant\Models\MultiVectorStruct;
use Qdrant\Models\PointStruct;
use Qdrant\Models\PointsStruct;

$vectors = new MultiVectorStruct();
$vectors->addVector('content',   [0.12, 0.53, /* ...1536 dims */]);
$vectors->addVector('thumbnail', [0.88, 0.12, /* ...512 dims  */]);

$points = new PointsStruct();
$points->addPoint(new PointStruct(
    1,
    $vectors,
    ['title' => 'My Article', 'published' => true]
));

$response = $client->collections('articles')->points()->upsert($points, ['wait' => 'true']);
```

--------------------------------

### Batch Query — `Query::batch()`

Source: https://context7.com/hkulekci/qdrant-php/llms.txt

Execute multiple independent queries in a single HTTP request to reduce network latency. Each query in the batch can have its own parameters, including limits and whether to return payloads or vectors.

```php
use Qdrant\Models\Request\Points\BatchQueryRequest;
use Qdrant\Models\Request\Points\QueryRequest;

$req1 = (new QueryRequest())
    ->setQuery(['nearest' => [0.1, 0.2, 0.3]])
    ->setUsing('image')
    ->setLimit(5);

$req2 = (new QueryRequest())
    ->setQuery(['nearest' => [0.7, 0.8, 0.9]])
    ->setUsing('image')
    ->setLimit(3)
    ->setWithPayload(true);

$batchRequest = new BatchQueryRequest([$req1, $req2]);
$response = $client->collections('products')->points()->query()->batch($batchRequest);

// $response['result'] is an array of result sets, one per query
foreach ($response['result'] as $i => $result) {
    echo "Query $i: " . count($result['points']) . " results" . PHP_EOL;
}
```

--------------------------------

### Create and Delete Field Indexes

Source: https://context7.com/hkulekci/qdrant-php/llms.txt

Create payload field indexes to accelerate filtered queries. Supports keyword, integer, float, full-text, and geo indexes.

```php
use Qdrant\Models\Request\CreateIndex;
use Qdrant\Models\Field\TextField;

// Keyword index (for exact match filtering)
$client->collections('articles')->index()->create(
    new CreateIndex('status', 'keyword')
);
```

```php
$client->collections('articles')->index()->create(
    new CreateIndex('year', 'integer')
);
```

```php
$client->collections('articles')->index()->create(
    new CreateIndex('price', 'float')
);
```

```php
// Full-text index with custom tokenizer
$textField = new TextField();
$client->collections('articles')->index()->create(
    new CreateIndex('body', $textField->schema())
);
```

```php
// Geo index
$client->collections('articles')->index()->create(
    new CreateIndex('location', 'geo')
);
```

```php
// Delete an index
$client->collections('articles')->index()->delete('status');
```

--------------------------------

### Field Index

Source: https://context7.com/hkulekci/qdrant-php/llms.txt

Create payload field indexes to accelerate filtered queries using `Index::create()` and remove them with `Index::delete()`.

```APIDOC
## Field Index — `Index::create()`, `Index::delete()`

Create payload field indexes to accelerate filtered queries.

```php
use Qdrant\Models\Request\CreateIndex;
use Qdrant\Models\Field\TextField;

// Keyword index (for exact match filtering)
$client->collections('articles')->index()->create(
    new CreateIndex('status', 'keyword')
);

// Integer index
$client->collections('articles')->index()->create(
    new CreateIndex('year', 'integer')
);

// Float index
$client->collections('articles')->index()->create(
    new CreateIndex('price', 'float')
);

// Full-text index with custom tokenizer
$textField = new TextField();
$client->collections('articles')->index()->create(
    new CreateIndex('body', $textField->schema())
);

// Geo index
$client->collections('articles')->index()->create(
    new CreateIndex('location', 'geo')
);

// Delete an index
$client->collections('articles')->index()->delete('status');
```
```

--------------------------------

### Multi-Vector Points

Source: https://context7.com/hkulekci/qdrant-php/llms.txt

Insert points into collections that have multiple named vector spaces.

```APIDOC
## Multi-Vector Points — `MultiVectorStruct`

Insert points into collections that have multiple named vector spaces.

```php
use Qdrant\Models\MultiVectorStruct;
use Qdrant\Models\PointStruct;
use Qdrant\Models\PointsStruct;

$vectors = new MultiVectorStruct();
$vectors->addVector('content',   [0.12, 0.53, /* ...1536 dims */]);
$vectors->addVector('thumbnail', [0.88, 0.12, /* ...512 dims  */]);

$points = new PointsStruct();
$points->addPoint(new PointStruct(
    1,
    $vectors,
    ['title' => 'My Article', 'published' => true]
));

$response = $client->collections('articles')->points()->upsert($points, ['wait' => 'true']);
```
```

--------------------------------

### Upsert Points with `Points::upsert()`

Source: https://context7.com/hkulekci/qdrant-php/llms.txt

Use this method to insert or update individual points with vectors and payloads. Supports synchronous acknowledgment by setting `wait` to 'true'. Ensure all necessary models are imported.

```php
use Qdrant\Models\PointsStruct;
use Qdrant\Models\PointStruct;
use Qdrant\Models\VectorStruct;

$points = new PointsStruct();
$points->addPoint(new PointStruct(
    1,                                               // integer or UUID string ID
    new VectorStruct([0.12, 0.53, /* ...1536 dims */], 'content'),
    ['title' => 'Sustainable Farming', 'year' => 2024, 'tags' => ['agri', 'eco']]
));
$points->addPoint(new PointStruct(
    2,
    new VectorStruct([0.88, 0.12, /* ... */], 'content'),
    ['title' => 'Urban Beekeeping', 'year' => 2023, 'tags' => ['urban', 'eco']]
));

// Upsert with wait=true to block until indexing is complete
$response = $client->collections('articles')->points()->upsert($points, ['wait' => 'true']);
// $response['status'] === 'ok'
// $response['result']['status'] === 'completed'
```

--------------------------------

### Scroll Points — `Points::scroll()`

Source: https://context7.com/hkulekci/qdrant-php/llms.txt

Iterate through all points in a collection, optionally filtered, without needing a query vector. Supports pagination using an offset or a cursor for efficient retrieval of large datasets. Can return payloads and vectors.

```php
use Qdrant\Models\Request\ScrollRequest;
use Qdrant\Models\Filter\Filter;
use Qdrant\Models\Filter\Condition\Range;

$scrollRequest = (new ScrollRequest())
    ->setFilter(
        (new Filter())->addMust(new Range('year', ['gte' => 2022, 'lte' => 2024]))
    )
    ->setLimit(100)
    ->setOffset(0)          // or pass a point ID string for cursor-based pagination
    ->setWithPayload(true)
    ->setWithVector(false);

$response = $client->collections('articles')->points()->scroll($scrollRequest);
$nextOffset = $response['result']['next_page_offset'];  // null if no more pages
foreach ($response['result']['points'] as $point) {
    echo $point['id'] . ': ' . $point['payload']['title'] . PHP_EOL;
}
```

--------------------------------

### Batch Upsert Points

Source: https://context7.com/hkulekci/qdrant-php/llms.txt

Efficiently batch-inserts points using the columnar batch format.

```APIDOC
## Batch Upsert Points — `Points::batch()`

Efficiently batch-insert points using the columnar batch format.

```php
use Qdrant\Models\Request\PointsBatch;
use Qdrant\Models\PointStruct;
use Qdrant\Models\VectorStruct;

$batch = new PointsBatch();
$batch->addPoint(PointStruct::createFromArray([
    'id'      => 1,
    'vector'  => new VectorStruct([0.1, 0.2, 0.3], 'image'),
    'payload' => ['color' => 'red', 'size' => 'large'],
]));
$batch->addPoint(PointStruct::createFromArray([
    'id'      => 2,
    'vector'  => new VectorStruct([0.4, 0.5, 0.6], 'image'),
    'payload' => ['color' => 'blue', 'size' => 'small'],
]));
$batch->addPoint(PointStruct::createFromArray([
    'id'      => 3,
    'vector'  => new VectorStruct([0.7, 0.8, 0.9], 'image'),
    'payload' => ['color' => 'green', 'size' => 'medium'],
]));

$response = $client->collections('products')->points()->batch($batch, ['wait' => 'true']);
// $response['status'] === 'ok'
```
```

--------------------------------

### Build Complex Filters with Conditions

Source: https://context7.com/hkulekci/qdrant-php/llms.txt

Compose `Filter` objects using `addMust()`, `addShould()`, and `addMustNot()` to define query conditions. Supports various data types and nested structures.

```php
use Qdrant\Models\Filter\Filter;
use Qdrant\Models\Filter\Nested;
use Qdrant\Models\Filter\Condition\MatchString;
use Qdrant\Models\Filter\Condition\MatchInt;
use Qdrant\Models\Filter\Condition\MatchBool;
use Qdrant\Models\Filter\Condition\MatchAny;
use Qdrant\Models\Filter\Condition\MatchExcept;
use Qdrant\Models\Filter\Condition\Range;
use Qdrant\Models\Filter\Condition\HasId;
use Qdrant\Models\Filter\Condition\IsEmpty;
use Qdrant\Models\Filter\Condition\FullTextMatch;
use Qdrant\Models\Filter\Condition\GeoRadius;
use Qdrant\Models\Filter\Condition\GeoBoundingBox;
use Qdrant\Models\Filter\Condition\ValueCount;

$filter = (new Filter())
    ->addMust(new MatchString('status', 'active'))            // exact string match
    ->addMust(new MatchInt('category_id', 7))                 // exact int match
    ->addMust(new MatchBool('is_featured', true))             // boolean match
    ->addMust(new MatchAny('tags', ['eco', 'bio', 'green']))  // match any of list
    ->addMust(new MatchExcept('region', ['banned', 'test']))  // exclude these values
    ->addMust(new Range('price', ['gte' => 10.0, 'lte' => 500.0]))
    ->addMust(new HasId([1, 5, 99]))                          // restrict to IDs
    ->addMustNot(new IsEmpty('description'))                  // field must not be empty
    ->addMust(new FullTextMatch('body', 'organic farming'))   // full-text search
    ->addMust(new GeoRadius('location', [
        'center' => ['lat' => 52.52, 'lon' => 13.405],
        'radius' => 5000,                                     // metres
    ]))
    ->addMust(new GeoBoundingBox('location', [
        'top_left'     => ['lat' => 53.0, 'lon' => 13.0],
        'bottom_right' => ['lat' => 52.0, 'lon' => 14.0],
    ]))
    ->addMust(new ValueCount('comments', ['gte' => 1]));      // array field count

// Nested filter (for object array fields)
$nestedFilter = new Filter();
$nestedFilter->addMust(new MatchString('color', 'red'));
$filter->addMust(new Nested('attributes', $nestedFilter));
```

--------------------------------

### Retrieve Points by ID — `Points::id()`, `Points::ids()`

Source: https://context7.com/hkulekci/qdrant-php/llms.txt

Fetch one or multiple points directly from a collection using their unique IDs. This method is efficient for retrieving specific known points. You can specify whether to include payloads and vectors in the response.

```php
// Single point
$response = $client->collections('articles')->points()->id(42);
echo $response['result']['id'];
echo $response['result']['payload']['title'];

// Multiple points
$response = $client->collections('articles')->points()->ids(
    [1, 2, 3, 42],
    withPayload: true,
    withVector: false
);
foreach ($response['result'] as $point) {
    echo $point['id'] . ': ' . $point['payload']['title'] . PHP_EOL;
}
```

--------------------------------

### Update Collection Parameters

Source: https://context7.com/hkulekci/qdrant-php/llms.txt

Allows modification of optimizer settings, HNSW parameters, collection parameters, or quantization on an existing collection.

```APIDOC
## Update Collection Parameters — `Collections::update()`

Modify optimizer settings, HNSW parameters, collection params, or quantization on an existing collection without recreation.

```php
use Qdrant\Models\Request\UpdateCollection;
use Qdrant\Models\Request\CollectionConfig\OptimizersConfig;
use Qdrant\Models\Request\CollectionConfig\CollectionParams;

$update = new UpdateCollection();
$update->setOptimizersConfig(
    (new OptimizersConfig())
        ->setIndexingThreshold(20000)
        ->setDefaultSegmentNumber(4)
        ->setFlushIntervalSec(5)
);
$update->setCollectionParams(
    (new CollectionParams())
        ->setReplicationFactor(2)
        ->setWriteConsistencyFactor(1)
        ->setOnDiskPayload(true)
);

$response = $client->collections('articles')->update($update);
// $response['result'] === true
```
```

--------------------------------

### Batch Query

Source: https://context7.com/hkulekci/qdrant-php/llms.txt

Execute multiple independent queries in a single HTTP request to optimize performance by reducing network round-trips. This is useful when you need to perform several distinct searches simultaneously.

```APIDOC
## Batch Query — `Query::batch()`

Run multiple independent queries in a single HTTP request to minimize round-trips.

```php
use Qdrant\Models\Request\Points\BatchQueryRequest;
use Qdrant\Models\Request\Points\QueryRequest;

$req1 = (new QueryRequest())
    ->setQuery(['nearest' => [0.1, 0.2, 0.3]])
    ->setUsing('image')
    ->setLimit(5);

$req2 = (new QueryRequest())
    ->setQuery(['nearest' => [0.7, 0.8, 0.9]])
    ->setUsing('image')
    ->setLimit(3)
    ->setWithPayload(true);

$batchRequest = new BatchQueryRequest([$req1, $req2]);
$response = $client->collections('products')->points()->query()->batch($batchRequest);

// $response['result'] is an array of result sets, one per query
foreach ($response['result'] as $i => $result) {
    echo "Query $i: " . count($result['points']) . " results" . PHP_EOL;
}
```
```

--------------------------------

### Update Collection Parameters

Source: https://context7.com/hkulekci/qdrant-php/llms.txt

Modify existing collection parameters such as optimizer settings, HNSW configuration, replication factor, and write consistency without recreating the collection.

```php
use Qdrant\Models\Request\UpdateCollection;
use Qdrant\Models\Request\CollectionConfig\OptimizersConfig;
use Qdrant\Models\Request\CollectionConfig\CollectionParams;

$update = new UpdateCollection();
$update->setOptimizersConfig(
    (new OptimizersConfig())
        ->setIndexingThreshold(20000)
        ->setDefaultSegmentNumber(4)
        ->setFlushIntervalSec(5)
);
$update->setCollectionParams(
    (new CollectionParams())
        ->setReplicationFactor(2)
        ->setWriteConsistencyFactor(1)
        ->setOnDiskPayload(true)
);

$response = $client->collections('articles')->update($update);
// $response['result'] === true
```