Performance

Learn how Redis vector sets behave under load and how to optimize for speed and recall

Vector set is a new data type that is currently in preview and may be subject to change.

Query performance

Vector similarity queries using the VSIM are threaded by default. Redis uses up to 32 threads to process these queries in parallel.

VSIM performance scales nearly linearly with available CPU cores.
Expect ~50,000 similarity queries per second for a 3M-item set with 300-dim vectors using int8 quantization.
Performance depends heavily on the EF parameter:
- Higher EF improves recall, but slows down search.
- Lower EF returns faster results with reduced accuracy.

Insertion performance

Inserting vectors with the VADD command is more computationally expensive than querying:

Insertion is single-threaded by default.
Use the CAS option to offload candidate graph search to a background thread.
Expect a few thousand insertions per second on a single node.

Quantization effects

Quantization greatly impacts both speed and memory:

Q8 (default): 4x smaller than FP32, high recall, high speed
BIN (binary): 32x smaller than FP32, lower recall, fastest search
NOQUANT (FP32): Full precision, slower performance, highest memory use

Use the quantization mode that best fits your tradeoff between precision and efficiency. The examples below show how the different modes affect a simple vector. Note that even with NOQUANT mode, the values change slightly, due to floating point rounding.

>_ Redis CLI

> VADD quantSetQ8 VALUES 2 1.262185 1.958231 quantElement Q8
(integer) 1
> VEMB quantSetQ8 quantElement
1) "1.2643694877624512"
2) "1.958230972290039"

> VADD quantSetNoQ VALUES 2 1.262185 1.958231 quantElement NOQUANT
(integer) 1
> VEMB quantSetNoQ quantElement
1) "1.262184977531433"
2) "1.958230972290039"

> VADD quantSetBin VALUES 2 1.262185 1.958231 quantElement BIN
(integer) 1
> VEMB quantSetBin quantElement
1) "1"
2) "1"

Python

"""
Code samples for Vector set doc pages:
    https://redis.io/docs/latest/develop/data-types/vector-sets/
"""

import redis

from redis.commands.vectorset.commands import (
    QuantizationOptions
)

r = redis.Redis(decode_responses=True)


res1 = r.vset().vadd("points", [1.0, 1.0], "pt:A")
print(res1)  # >>> 1

res2 = r.vset().vadd("points", [-1.0, -1.0], "pt:B")
print(res2)  # >>> 1

res3 = r.vset().vadd("points", [-1.0, 1.0], "pt:C")
print(res3)  # >>> 1

res4 = r.vset().vadd("points", [1.0, -1.0], "pt:D")
print(res4)  # >>> 1

res5 = r.vset().vadd("points", [1.0, 0], "pt:E")
print(res5)  # >>> 1

res6 = r.type("points")
print(res6)  # >>> vectorset

res7 = r.vset().vcard("points")
print(res7)  # >>> 5

res8 = r.vset().vdim("points")
print(res8)  # >>> 2

res9 = r.vset().vemb("points", "pt:A")
print(res9)  # >>> [0.9999999403953552, 0.9999999403953552]

res10 = r.vset().vemb("points", "pt:B")
print(res10)  # >>> [-0.9999999403953552, -0.9999999403953552]

res11 = r.vset().vemb("points", "pt:C")
print(res11)  # >>> [-0.9999999403953552, 0.9999999403953552]

res12 = r.vset().vemb("points", "pt:D")
print(res12)  # >>> [0.9999999403953552, -0.9999999403953552]

res13 = r.vset().vemb("points", "pt:E")
print(res13)  # >>> [1, 0]

res14 = r.vset().vsetattr("points", "pt:A", {
    "name": "Point A",
    "description": "First point added"
})
print(res14)  # >>> 1

res15 = r.vset().vgetattr("points", "pt:A")
print(res15)
# >>> {'name': 'Point A', 'description': 'First point added'}

res16 = r.vset().vsetattr("points", "pt:A", "")
print(res16)  # >>> 1

res17 = r.vset().vgetattr("points", "pt:A")
print(res17)  # >>> None

res18 = r.vset().vadd("points", [0, 0], "pt:F")
print(res18)  # >>> 1

res19 = r.vset().vcard("points")
print(res19)  # >>> 6

res20 = r.vset().vrem("points", "pt:F")
print(res20)  # >>> 1

res21 = r.vset().vcard("points")
print(res21)  # >>> 5

res22 = r.vset().vsim("points", [0.9, 0.1])
print(res22)
# >>> ['pt:E', 'pt:A', 'pt:D', 'pt:C', 'pt:B']

res23 = r.vset().vsim(
    "points", "pt:A",
    with_scores=True,
    count=4
)
print(res23)
# >>> {'pt:A': 1.0, 'pt:E': 0.8535534143447876, 'pt:D': 0.5, 'pt:C': 0.5}

res24 = r.vset().vsetattr("points", "pt:A", {
    "size": "large",
    "price": 18.99
})
print(res24)  # >>> 1

res25 = r.vset().vsetattr("points", "pt:B", {
    "size": "large",
    "price": 35.99
})
print(res25)  # >>> 1

res26 = r.vset().vsetattr("points", "pt:C", {
    "size": "large",
    "price": 25.99
})
print(res26)  # >>> 1

res27 = r.vset().vsetattr("points", "pt:D", {
    "size": "small",
    "price": 21.00
})
print(res27)  # >>> 1

res28 = r.vset().vsetattr("points", "pt:E", {
    "size": "small",
    "price": 17.75
})
print(res28)  # >>> 1

# Return elements in order of distance from point A whose
# `size` attribute is `large`.
res29 = r.vset().vsim(
    "points", "pt:A",
    filter='.size == "large"'
)
print(res29)  # >>> ['pt:A', 'pt:C', 'pt:B']

# Return elements in order of distance from point A whose size is
# `large` and whose price is greater than 20.00.
res30 = r.vset().vsim(
    "points", "pt:A",
    filter='.size == "large" && .price > 20.00'
)
print(res30)  # >>> ['pt:C', 'pt:B']

# Import `QuantizationOptions` enum using:
#
# from redis.commands.vectorset.commands import (
#   QuantizationOptions
# )
res31 = r.vset().vadd(
    "quantSetQ8", [1.262185, 1.958231],
    "quantElement",
    quantization=QuantizationOptions.Q8
)
print(res31)  # >>> 1

res32 = r.vset().vemb("quantSetQ8", "quantElement")
print(f"Q8: {res32}")
# >>> Q8: [1.2643694877624512, 1.958230972290039]

res33 = r.vset().vadd(
    "quantSetNoQ", [1.262185, 1.958231],
    "quantElement",
    quantization=QuantizationOptions.NOQUANT
)
print(res33)  # >>> 1

res34 = r.vset().vemb("quantSetNoQ", "quantElement")
print(f"NOQUANT: {res34}")
# >>> NOQUANT: [1.262184977531433, 1.958230972290039]

res35 = r.vset().vadd(
    "quantSetBin", [1.262185, 1.958231],
    "quantElement",
    quantization=QuantizationOptions.BIN
)
print(res35)  # >>> 1

res36 = r.vset().vemb("quantSetBin", "quantElement")
print(f"BIN: {res36}")
# >>> BIN: [1, 1]

# Create a list of 300 arbitrary values.
values = [x / 299 for x in range(300)]

res37 = r.vset().vadd(
    "setNotReduced",
    values,
    "element"
)
print(res37)  # >>> 1

res38 = r.vset().vdim("setNotReduced")
print(res38)  # >>> 300

res39 = r.vset().vadd(
    "setReduced",
    values,
    "element",
    reduce_dim=100
)
print(res39)  # >>> 1

res40 = r.vset().vdim("setReduced")  # >>> 100
print(res40)

Java-Async

package io.redis.examples.async;

import io.lettuce.core.*;
import io.lettuce.core.api.async.RedisAsyncCommands;
import io.lettuce.core.api.StatefulRedisConnection;
import io.lettuce.core.vector.QuantizationType;


import java.util.concurrent.CompletableFuture;

public class VectorSetExample {

    public void run() {
        RedisClient redisClient = RedisClient.create("redis://localhost:6379");

        try (StatefulRedisConnection<String, String> connection = redisClient.connect()) {
            RedisAsyncCommands<String, String> asyncCommands = connection.async();


            CompletableFuture<Boolean> addPointA = asyncCommands.vadd("points", "pt:A", 1.0, 1.0).thenApply(result -> {
                System.out.println(result); // >>> true
                return result;
            }).toCompletableFuture();

            CompletableFuture<Boolean> addPointB = asyncCommands.vadd("points", "pt:B", -1.0, -1.0).thenApply(result -> {
                System.out.println(result); // >>> true
                return result;
            }).toCompletableFuture();

            CompletableFuture<Boolean> addPointC = asyncCommands.vadd("points", "pt:C", -1.0, 1.0).thenApply(result -> {
                System.out.println(result); // >>> true
                return result;
            }).toCompletableFuture();

            CompletableFuture<Boolean> addPointD = asyncCommands.vadd("points", "pt:D", 1.0, -1.0).thenApply(result -> {
                System.out.println(result); // >>> true
                return result;
            }).toCompletableFuture();

            CompletableFuture<Boolean> addPointE = asyncCommands.vadd("points", "pt:E", 1.0, 0.0).thenApply(result -> {
                System.out.println(result); // >>> true
                return result;
            }).toCompletableFuture();

            // Chain checkDataType after all vadd operations complete
            CompletableFuture<Void> vaddOperations = CompletableFuture
                    .allOf(addPointA, addPointB, addPointC, addPointD, addPointE)
                    .thenCompose(ignored -> asyncCommands.type("points")).thenAccept(result -> {
                        System.out.println(result); // >>> vectorset
                    }).toCompletableFuture();

            CompletableFuture<Void> getCardinality = asyncCommands.vcard("points").thenAccept(result -> {
                System.out.println(result); // >>> 5
            }).toCompletableFuture();

            CompletableFuture<Void> getDimensions = asyncCommands.vdim("points").thenAccept(result -> {
                System.out.println(result); // >>> 2
            }).toCompletableFuture();

            CompletableFuture<Void> getEmbeddingA = asyncCommands.vemb("points", "pt:A").thenAccept(result -> {
                System.out.println(result); // >>> [0.9999999403953552, 0.9999999403953552]
            }).toCompletableFuture();

            CompletableFuture<Void> getEmbeddingB = asyncCommands.vemb("points", "pt:B").thenAccept(result -> {
                System.out.println(result); // >>> [-0.9999999403953552, -0.9999999403953552]
            }).toCompletableFuture();

            CompletableFuture<Void> getEmbeddingC = asyncCommands.vemb("points", "pt:C").thenAccept(result -> {
                System.out.println(result); // >>> [-0.9999999403953552, 0.9999999403953552]
            }).toCompletableFuture();

            CompletableFuture<Void> getEmbeddingD = asyncCommands.vemb("points", "pt:D").thenAccept(result -> {
                System.out.println(result); // >>> [0.9999999403953552, -0.9999999403953552]
            }).toCompletableFuture();

            CompletableFuture<Void> getEmbeddingE = asyncCommands.vemb("points", "pt:E").thenAccept(result -> {
                System.out.println(result); // >>> [1.0, 0.0]
            }).toCompletableFuture();

            CompletableFuture<Void> setAttributeA = asyncCommands
                    .vsetattr("points", "pt:A", "{\"name\": \"Point A\", \"description\": \"First point added\"}")
                    .thenAccept(result -> {
                        System.out.println(result); // >>> true
                    }).toCompletableFuture();

            CompletableFuture<Void> getAttributeA = asyncCommands.vgetattr("points", "pt:A").thenAccept(result -> {
                System.out.println(result); // >>> {"name": "Point A", "description": "First point added"}
            }).toCompletableFuture();

            CompletableFuture<Void> clearAttributeA = asyncCommands.vsetattr("points", "pt:A", "").thenAccept(result -> {
                System.out.println(result); // >>> true
            }).toCompletableFuture();

            CompletableFuture<Void> verifyAttributeCleared = asyncCommands.vgetattr("points", "pt:A").thenAccept(result -> {
                System.out.println(result); // >>> null
            }).toCompletableFuture();

            CompletableFuture<Void> addTempPointF = asyncCommands.vadd("points", "pt:F", 0.0, 0.0).thenAccept(result -> {
                System.out.println(result); // >>> true
            }).toCompletableFuture();

            CompletableFuture<Void> checkCardinalityBefore = asyncCommands.vcard("points").thenAccept(result -> {
                System.out.println(result); // >>> 6
            }).toCompletableFuture();

            CompletableFuture<Void> removePointF = asyncCommands.vrem("points", "pt:F").thenAccept(result -> {
                System.out.println(result); // >>> true
            }).toCompletableFuture();

            CompletableFuture<Void> checkCardinalityAfter = asyncCommands.vcard("points").thenAccept(result -> {
                System.out.println(result); // >>> 5
            }).toCompletableFuture();

            CompletableFuture<Void> basicSimilaritySearch = asyncCommands.vsim("points", 0.9, 0.1).thenAccept(result -> {
                System.out.println(result); // >>> [pt:E, pt:A, pt:D, pt:C, pt:B]
            }).toCompletableFuture();

            VSimArgs vsimArgs = new VSimArgs();
            vsimArgs.count(4L);
            CompletableFuture<Void> similaritySearchWithScore = asyncCommands.vsimWithScore("points", vsimArgs, "pt:A")
                    .thenAccept(result -> {
                        System.out.println(result); // >>> {pt:A=1.0, pt:E=0.8535534143447876, pt:D=0.5, pt:C=0.5}
                    }).toCompletableFuture();

            CompletableFuture<Void> filteredSimilaritySearch = asyncCommands
                    .vsetattr("points", "pt:A", "{\"size\": \"large\", \"price\": 18.99}").thenCompose(result -> {
                        System.out.println(result); // >>> true
                        return asyncCommands.vsetattr("points", "pt:B", "{\"size\": \"large\", \"price\": 35.99}");
                    }).thenCompose(result -> {
                        System.out.println(result); // >>> true
                        return asyncCommands.vsetattr("points", "pt:C", "{\"size\": \"large\", \"price\": 25.99}");
                    }).thenCompose(result -> {
                        System.out.println(result); // >>> true
                        return asyncCommands.vsetattr("points", "pt:D", "{\"size\": \"small\", \"price\": 21.00}");
                    }).thenCompose(result -> {
                        System.out.println(result); // >>> true
                        return asyncCommands.vsetattr("points", "pt:E", "{\"size\": \"small\", \"price\": 17.75}");
                    }).thenCompose(result -> {
                        System.out.println(result); // >>> true

                        // Return elements in order of distance from point A whose size attribute is large.
                        VSimArgs filterArgs = new VSimArgs();
                        filterArgs.filter(".size == \"large\"");
                        return asyncCommands.vsim("points", filterArgs, "pt:A");
                    }).thenCompose(result -> {
                        System.out.println(result); // >>> [pt:A, pt:C, pt:B]

                        // Return elements in order of distance from point A whose size is large and price > 20.00.
                        VSimArgs filterArgs2 = new VSimArgs();
                        filterArgs2.filter(".size == \"large\" && .price > 20.00");
                        return asyncCommands.vsim("points", filterArgs2, "pt:A");
                    }).thenAccept(result -> {
                        System.out.println(result); // >>> [pt:C, pt:B]
                    }).toCompletableFuture();
            VAddArgs q8Args = VAddArgs.Builder.quantizationType(QuantizationType.Q8);
            CompletableFuture<Void> quantizationOperations = asyncCommands
                    .vadd("quantSetQ8", "quantElement", q8Args, 1.262185, 1.958231).thenCompose(result -> {
                        System.out.println(result); // >>> true
                        return asyncCommands.vemb("quantSetQ8", "quantElement");
                    }).thenCompose(result -> {
                        System.out.println("Q8: " + result); // >>> Q8: [1.2643694877624512, 1.958230972290039]

                        VAddArgs noQuantArgs = VAddArgs.Builder.quantizationType(QuantizationType.NO_QUANTIZATION);
                        return asyncCommands.vadd("quantSetNoQ", "quantElement", noQuantArgs, 1.262185, 1.958231);
                    }).thenCompose(result -> {
                        System.out.println(result); // >>> true
                        return asyncCommands.vemb("quantSetNoQ", "quantElement");
                    }).thenCompose(result -> {
                        System.out.println("NOQUANT: " + result); // >>> NOQUANT: [1.262184977531433, 1.958230972290039]

                        VAddArgs binArgs = VAddArgs.Builder.quantizationType(QuantizationType.BINARY);
                        return asyncCommands.vadd("quantSetBin", "quantElement", binArgs, 1.262185, 1.958231);
                    }).thenCompose(result -> {
                        System.out.println(result); // >>> true
                        return asyncCommands.vemb("quantSetBin", "quantElement");
                    }).thenAccept(result -> {
                        System.out.println("BIN: " + result); // >>> BIN: [1.0, 1.0]
                    }).toCompletableFuture();

            // Create a list of 300 arbitrary values.
            Double[] values = new Double[300];
            for (int i = 0; i < 300; i++) {
                values[i] = (double) i / 299;
            }

            CompletableFuture<Void> dimensionalityReductionOperations = asyncCommands.vadd("setNotReduced", "element", values)
                    .thenCompose(result -> {
                        System.out.println(result); // >>> true
                        return asyncCommands.vdim("setNotReduced");
                    }).thenCompose(result -> {
                        System.out.println(result); // >>> 300
                        return asyncCommands.vadd("setReduced", 100, "element", values);
                    }).thenCompose(result -> {
                        System.out.println(result); // >>> true
                        return asyncCommands.vdim("setReduced");
                    }).thenAccept(result -> {
                        System.out.println(result); // >>> 100
                    }).toCompletableFuture();

            // Wait for all async operations to complete
            CompletableFuture.allOf(
                    // Vector addition operations (chained: parallel vadd + sequential checkDataType)
                    vaddOperations,
                    // Cardinality and dimension operations
                    getCardinality, getDimensions,
                    // Vector embedding retrieval operations
                    getEmbeddingA, getEmbeddingB, getEmbeddingC, getEmbeddingD, getEmbeddingE,
                    // Attribute operations
                    setAttributeA, getAttributeA, clearAttributeA, verifyAttributeCleared,
                    // Vector removal operations
                    addTempPointF, checkCardinalityBefore, removePointF, checkCardinalityAfter,
                    // Similarity search operations
                    basicSimilaritySearch, similaritySearchWithScore, filteredSimilaritySearch,
                    // Advanced operations
                    quantizationOperations, dimensionalityReductionOperations).join();

        } finally {
            redisClient.shutdown();
        }
    }

}

Java-Reactive

package io.redis.examples.reactive;

import io.lettuce.core.*;
import io.lettuce.core.api.reactive.RedisReactiveCommands;
import io.lettuce.core.api.StatefulRedisConnection;
import io.lettuce.core.vector.QuantizationType;


import reactor.core.publisher.Mono;

public class VectorSetExample {

    public void run() {
        RedisClient redisClient = RedisClient.create("redis://localhost:6379");

        try (StatefulRedisConnection<String, String> connection = redisClient.connect()) {
            RedisReactiveCommands<String, String> reactiveCommands = connection.reactive();


            Mono<Boolean> addPointA = reactiveCommands.vadd("points", "pt:A", 1.0, 1.0).doOnNext(result -> {
                System.out.println(result); // >>> true
            });

            Mono<Boolean> addPointB = reactiveCommands.vadd("points", "pt:B", -1.0, -1.0).doOnNext(result -> {
                System.out.println(result); // >>> true
            });

            Mono<Boolean> addPointC = reactiveCommands.vadd("points", "pt:C", -1.0, 1.0).doOnNext(result -> {
                System.out.println(result); // >>> true
            });

            Mono<Boolean> addPointD = reactiveCommands.vadd("points", "pt:D", 1.0, -1.0).doOnNext(result -> {
                System.out.println(result); // >>> true
            });

            Mono<Boolean> addPointE = reactiveCommands.vadd("points", "pt:E", 1.0, 0.0).doOnNext(result -> {
                System.out.println(result); // >>> true
            });

            Mono<Void> vaddOperations = Mono.when(addPointA, addPointB, addPointC, addPointD, addPointE)
                    .then(reactiveCommands.type("points")).doOnNext(result -> {
                        System.out.println(result); // >>> vectorset
                    }).then();

            Mono<Long> getCardinality = reactiveCommands.vcard("points").doOnNext(result -> {
                System.out.println(result); // >>> 5
            });

            Mono<Long> getDimensions = reactiveCommands.vdim("points").doOnNext(result -> {
                System.out.println(result); // >>> 2
            });

            Mono<java.util.List<Double>> getEmbeddingA = reactiveCommands.vemb("points", "pt:A").collectList()
                    .doOnNext(result -> {
                        System.out.println(result); // >>> [0.9999999403953552, 0.9999999403953552]
                    });

            Mono<java.util.List<Double>> getEmbeddingB = reactiveCommands.vemb("points", "pt:B").collectList()
                    .doOnNext(result -> {
                        System.out.println(result); // >>> [-0.9999999403953552, -0.9999999403953552]
                    });

            Mono<java.util.List<Double>> getEmbeddingC = reactiveCommands.vemb("points", "pt:C").collectList()
                    .doOnNext(result -> {
                        System.out.println(result); // >>> [-0.9999999403953552, 0.9999999403953552]
                    });

            Mono<java.util.List<Double>> getEmbeddingD = reactiveCommands.vemb("points", "pt:D").collectList()
                    .doOnNext(result -> {
                        System.out.println(result); // >>> [0.9999999403953552, -0.9999999403953552]
                    });

            Mono<java.util.List<Double>> getEmbeddingE = reactiveCommands.vemb("points", "pt:E").collectList()
                    .doOnNext(result -> {
                        System.out.println(result); // >>> [1.0, 0.0]
                    });

            Mono<Boolean> setAttributeA = reactiveCommands
                    .vsetattr("points", "pt:A", "{\"name\": \"Point A\", \"description\": \"First point added\"}")
                    .doOnNext(result -> {
                        System.out.println(result); // >>> true
                    });

            Mono<String> getAttributeA = reactiveCommands.vgetattr("points", "pt:A").doOnNext(result -> {
                System.out.println(result); // >>> {"name": "Point A", "description": "First point added"}
            });

            Mono<Boolean> clearAttributeA = reactiveCommands.vsetattr("points", "pt:A", "").doOnNext(result -> {
                System.out.println(result); // >>> true
            });

            Mono<String> verifyAttributeCleared = reactiveCommands.vgetattr("points", "pt:A").doOnNext(result -> {
                System.out.println(result); // >>> null
            });

            Mono<Boolean> addTempPointF = reactiveCommands.vadd("points", "pt:F", 0.0, 0.0).doOnNext(result -> {
                System.out.println(result); // >>> true
            });

            Mono<Long> checkCardinalityBefore = reactiveCommands.vcard("points").doOnNext(result -> {
                System.out.println(result); // >>> 6
            });

            Mono<Boolean> removePointF = reactiveCommands.vrem("points", "pt:F").doOnNext(result -> {
                System.out.println(result); // >>> true
            });

            Mono<Long> checkCardinalityAfter = reactiveCommands.vcard("points").doOnNext(result -> {
                System.out.println(result); // >>> 5
            });

            Mono<java.util.List<String>> basicSimilaritySearch = reactiveCommands.vsim("points", 0.9, 0.1).collectList()
                    .doOnNext(result -> {
                        System.out.println(result); // >>> [pt:E, pt:A, pt:D, pt:C, pt:B]
                    });

            VSimArgs vsimArgs = new VSimArgs();
            vsimArgs.count(4L);
            Mono<java.util.Map<String, Double>> similaritySearchWithScore = reactiveCommands
                    .vsimWithScore("points", vsimArgs, "pt:A").doOnNext(result -> {
                        System.out.println(result); // >>> {pt:A=1.0, pt:E=0.8535534143447876, pt:D=0.5, pt:C=0.5}
                    });

            Mono<Void> filteredSimilaritySearch = reactiveCommands
                    .vsetattr("points", "pt:A", "{\"size\": \"large\", \"price\": 18.99}").doOnNext(result -> {
                        System.out.println(result); // >>> true
                    }).flatMap(result -> reactiveCommands.vsetattr("points", "pt:B", "{\"size\": \"large\", \"price\": 35.99}"))
                    .doOnNext(result -> {
                        System.out.println(result); // >>> true
                    }).flatMap(result -> reactiveCommands.vsetattr("points", "pt:C", "{\"size\": \"large\", \"price\": 25.99}"))
                    .doOnNext(result -> {
                        System.out.println(result); // >>> true
                    }).flatMap(result -> reactiveCommands.vsetattr("points", "pt:D", "{\"size\": \"small\", \"price\": 21.00}"))
                    .doOnNext(result -> {
                        System.out.println(result); // >>> true
                    }).flatMap(result -> reactiveCommands.vsetattr("points", "pt:E", "{\"size\": \"small\", \"price\": 17.75}"))
                    .doOnNext(result -> {
                        System.out.println(result); // >>> true
                    }).flatMap(result -> {
                        // Return elements in order of distance from point A whose size attribute is large.
                        VSimArgs filterArgs = new VSimArgs();
                        filterArgs.filter(".size == \"large\"");
                        return reactiveCommands.vsim("points", filterArgs, "pt:A").collectList();
                    }).doOnNext(result -> {
                        System.out.println(result); // >>> [pt:A, pt:C, pt:B]
                    }).flatMap(result -> {
                        // Return elements in order of distance from point A whose size is large and price > 20.00.
                        VSimArgs filterArgs2 = new VSimArgs();
                        filterArgs2.filter(".size == \"large\" && .price > 20.00");
                        return reactiveCommands.vsim("points", filterArgs2, "pt:A").collectList();
                    }).doOnNext(result -> {
                        System.out.println(result); // >>> [pt:C, pt:B]
                    }).then();

            VAddArgs q8Args = VAddArgs.Builder.quantizationType(QuantizationType.Q8);
            Mono<Void> quantizationOperations = reactiveCommands.vadd("quantSetQ8", "quantElement", q8Args, 1.262185, 1.958231)
                    .doOnNext(result -> {
                        System.out.println(result); // >>> true
                    }).flatMap(result -> reactiveCommands.vemb("quantSetQ8", "quantElement").collectList()).doOnNext(result -> {
                        System.out.println("Q8: " + result); // >>> Q8: [1.2643694877624512, 1.958230972290039]
                    }).flatMap(result -> {
                        VAddArgs noQuantArgs = VAddArgs.Builder.quantizationType(QuantizationType.NO_QUANTIZATION);
                        return reactiveCommands.vadd("quantSetNoQ", "quantElement", noQuantArgs, 1.262185, 1.958231);
                    }).doOnNext(result -> {
                        System.out.println(result); // >>> true
                    }).flatMap(result -> reactiveCommands.vemb("quantSetNoQ", "quantElement").collectList())
                    .doOnNext(result -> {
                        System.out.println("NOQUANT: " + result); // >>> NOQUANT: [1.262184977531433, 1.958230972290039]
                    }).flatMap(result -> {
                        VAddArgs binArgs = VAddArgs.Builder.quantizationType(QuantizationType.BINARY);
                        return reactiveCommands.vadd("quantSetBin", "quantElement", binArgs, 1.262185, 1.958231);
                    }).doOnNext(result -> {
                        System.out.println(result); // >>> true
                    }).flatMap(result -> reactiveCommands.vemb("quantSetBin", "quantElement").collectList())
                    .doOnNext(result -> {
                        System.out.println("BIN: " + result); // >>> BIN: [1.0, 1.0]
                    }).then();

            // Create a list of 300 arbitrary values.
            Double[] values = new Double[300];
            for (int i = 0; i < 300; i++) {
                values[i] = (double) i / 299;
            }

            Mono<Void> dimensionalityReductionOperations = reactiveCommands.vadd("setNotReduced", "element", values)
                    .doOnNext(result -> {
                        System.out.println(result); // >>> true
                    }).flatMap(result -> reactiveCommands.vdim("setNotReduced")).doOnNext(result -> {
                        System.out.println(result); // >>> 300
                    }).flatMap(result -> reactiveCommands.vadd("setReduced", 100, "element", values)).doOnNext(result -> {
                        System.out.println(result); // >>> true
                    }).flatMap(result -> reactiveCommands.vdim("setReduced")).doOnNext(result -> {
                        System.out.println(result); // >>> 100
                    }).then();

            // Wait for all reactive operations to complete
            Mono.when(
                    // Vector addition operations (chained sequentially)
                    vaddOperations,
                    // Cardinality and dimension operations
                    getCardinality, getDimensions,
                    // Vector embedding retrieval operations
                    getEmbeddingA, getEmbeddingB, getEmbeddingC, getEmbeddingD, getEmbeddingE,
                    // Attribute operations
                    setAttributeA, getAttributeA, clearAttributeA, verifyAttributeCleared,
                    // Vector removal operations
                    addTempPointF, checkCardinalityBefore, removePointF, checkCardinalityAfter,
                    // Similarity search operations
                    basicSimilaritySearch, similaritySearchWithScore, filteredSimilaritySearch,
                    // Advanced operations
                    quantizationOperations, dimensionalityReductionOperations).block();

        } finally {
            redisClient.shutdown();
        }
    }

}

package example_commands_test

import (
	"context"
	"fmt"
	"sort"

	"github.com/redis/go-redis/v9"
)


func ExampleClient_vectorset() {
	ctx := context.Background()

	rdb := redis.NewClient(&redis.Options{
		Addr:     "localhost:6379",
		Password: "", // no password set
		DB:       0,  // use default DB
	})

	defer rdb.Close()

	res1, err := rdb.VAdd(ctx, "points", "pt:A",
		&redis.VectorValues{Val: []float64{1.0, 1.0}},
	).Result()

	if err != nil {
		panic(err)
	}

	fmt.Println(res1) // >>> true

	res2, err := rdb.VAdd(ctx, "points", "pt:B",
		&redis.VectorValues{Val: []float64{-1.0, -1.0}},
	).Result()

	if err != nil {
		panic(err)
	}

	fmt.Println(res2) // >>> true

	res3, err := rdb.VAdd(ctx, "points", "pt:C",
		&redis.VectorValues{Val: []float64{-1.0, 1.0}},
	).Result()

	if err != nil {
		panic(err)
	}

	fmt.Println(res3) // >>> true

	res4, err := rdb.VAdd(ctx, "points", "pt:D",
		&redis.VectorValues{Val: []float64{1.0, -1.0}},
	).Result()

	if err != nil {
		panic(err)
	}

	fmt.Println(res4) // >>> true

	res5, err := rdb.VAdd(ctx, "points", "pt:E",
		&redis.VectorValues{Val: []float64{1.0, 0.0}},
	).Result()

	if err != nil {
		panic(err)
	}

	fmt.Println(res5) // >>> true

	res6, err := rdb.Type(ctx, "points").Result()

	if err != nil {
		panic(err)
	}

	fmt.Println(res6) // >>> vectorset

	res7, err := rdb.VCard(ctx, "points").Result()

	if err != nil {
		panic(err)
	}

	fmt.Println(res7) // >>> 5

	res8, err := rdb.VDim(ctx, "points").Result()

	if err != nil {
		panic(err)
	}

	fmt.Println(res8) // >>> 2

	res9, err := rdb.VEmb(ctx, "points", "pt:A", false).Result()

	if err != nil {
		panic(err)
	}

	fmt.Println(res9) // >>> [0.9999999403953552 0.9999999403953552]

	res10, err := rdb.VEmb(ctx, "points", "pt:B", false).Result()

	if err != nil {
		panic(err)
	}

	fmt.Println(res10) // >>> [-0.9999999403953552 -0.9999999403953552]

	res11, err := rdb.VEmb(ctx, "points", "pt:C", false).Result()

	if err != nil {
		panic(err)
	}

	fmt.Println(res11) // >>> [-0.9999999403953552 0.9999999403953552]

	res12, err := rdb.VEmb(ctx, "points", "pt:D", false).Result()

	if err != nil {
		panic(err)
	}

	fmt.Println(res12) // >>> [0.9999999403953552 -0.9999999403953552]

	res13, err := rdb.VEmb(ctx, "points", "pt:E", false).Result()

	if err != nil {
		panic(err)
	}

	fmt.Println(res13) // >>> [1 0]

	attrs := map[string]interface{}{
		"name":        "Point A",
		"description": "First point added",
	}

	res14, err := rdb.VSetAttr(ctx, "points", "pt:A", attrs).Result()

	if err != nil {
		panic(err)
	}

	fmt.Println(res14) // >>> true

	res15, err := rdb.VGetAttr(ctx, "points", "pt:A").Result()

	if err != nil {
		panic(err)
	}

	fmt.Println(res15)
	// >>> {"description":"First point added","name":"Point A"}

	res16, err := rdb.VClearAttributes(ctx, "points", "pt:A").Result()

	if err != nil {
		panic(err)
	}

	fmt.Println(res16) // >>> true

	// `VGetAttr()` returns an error if the attribute doesn't exist.
	_, err = rdb.VGetAttr(ctx, "points", "pt:A").Result()

	if err != nil {
		fmt.Println(err)
	}

	res18, err := rdb.VAdd(ctx, "points", "pt:F",
		&redis.VectorValues{Val: []float64{0.0, 0.0}},
	).Result()

	if err != nil {
		panic(err)
	}

	fmt.Println(res18) // >>> true

	res19, err := rdb.VCard(ctx, "points").Result()

	if err != nil {
		panic(err)
	}

	fmt.Println(res19) // >>> 6

	res20, err := rdb.VRem(ctx, "points", "pt:F").Result()

	if err != nil {
		panic(err)
	}

	fmt.Println(res20) // >>> true

	res21, err := rdb.VCard(ctx, "points").Result()

	if err != nil {
		panic(err)
	}

	fmt.Println(res21) // >>> 5

	res22, err := rdb.VSim(ctx, "points",
		&redis.VectorValues{Val: []float64{0.9, 0.1}},
	).Result()

	if err != nil {
		panic(err)
	}

	fmt.Println(res22) // >>> [pt:E pt:A pt:D pt:C pt:B]

	res23, err := rdb.VSimWithArgsWithScores(
		ctx,
		"points",
		&redis.VectorRef{Name: "pt:A"},
		&redis.VSimArgs{Count: 4},
	).Result()

	if err != nil {
		panic(err)
	}

	sort.Slice(res23, func(i, j int) bool {
		return res23[i].Name < res23[j].Name
	})

	fmt.Println(res23)
	// >>> [{pt:A 1} {pt:C 0.5} {pt:D 0.5} {pt:E 0.8535534143447876}]

	// Set attributes for filtering
	res24, err := rdb.VSetAttr(ctx, "points", "pt:A",
		map[string]interface{}{
			"size":  "large",
			"price": 18.99,
		},
	).Result()

	if err != nil {
		panic(err)
	}

	fmt.Println(res24) // >>> true

	res25, err := rdb.VSetAttr(ctx, "points", "pt:B",
		map[string]interface{}{
			"size":  "large",
			"price": 35.99,
		},
	).Result()

	if err != nil {
		panic(err)
	}

	fmt.Println(res25) // >>> true

	res26, err := rdb.VSetAttr(ctx, "points", "pt:C",
		map[string]interface{}{
			"size":  "large",
			"price": 25.99,
		},
	).Result()

	if err != nil {
		panic(err)
	}

	fmt.Println(res26) // >>> true

	res27, err := rdb.VSetAttr(ctx, "points", "pt:D",
		map[string]interface{}{
			"size":  "small",
			"price": 21.00,
		},
	).Result()

	if err != nil {
		panic(err)
	}

	fmt.Println(res27) // >>> true

	res28, err := rdb.VSetAttr(ctx, "points", "pt:E",
		map[string]interface{}{
			"size":  "small",
			"price": 17.75,
		},
	).Result()

	if err != nil {
		panic(err)
	}

	fmt.Println(res28) // >>> true

	// Return elements in order of distance from point A whose
	// `size` attribute is `large`.
	res29, err := rdb.VSimWithArgs(ctx, "points",
		&redis.VectorRef{Name: "pt:A"},
		&redis.VSimArgs{Filter: `.size == "large"`},
	).Result()

	if err != nil {
		panic(err)
	}

	fmt.Println(res29) // >>> [pt:A pt:C pt:B]

	// Return elements in order of distance from point A whose size is
	// `large` and whose price is greater than 20.00.
	res30, err := rdb.VSimWithArgs(ctx, "points",
		&redis.VectorRef{Name: "pt:A"},
		&redis.VSimArgs{Filter: `.size == "large" && .price > 20.00`},
	).Result()

	if err != nil {
		panic(err)
	}

	fmt.Println(res30) // >>> [pt:C pt:B]

}

func ExampleClient_vectorset_quantization() {
	ctx := context.Background()

	rdb := redis.NewClient(&redis.Options{
		Addr:     "localhost:6379",
		Password: "", // no password set
		DB:       0,  // use default DB
	})

	defer rdb.Close()

	// Add with Q8 quantization
	vecQ := &redis.VectorValues{Val: []float64{1.262185, 1.958231}}

	res1, err := rdb.VAddWithArgs(ctx, "quantSetQ8", "quantElement", vecQ,
		&redis.VAddArgs{
			Q8: true,
		},
	).Result()

	if err != nil {
		panic(err)
	}

	fmt.Println(res1) // >>> true

	embQ8, err := rdb.VEmb(ctx, "quantSetQ8", "quantElement", false).Result()

	if err != nil {
		panic(err)
	}

	fmt.Printf("Q8 embedding: %v\n", embQ8)
	// >>> Q8 embedding: [1.2621850967407227 1.9582309722900391]

	// Add with NOQUANT option
	res2, err := rdb.VAddWithArgs(ctx, "quantSetNoQ", "quantElement", vecQ,
		&redis.VAddArgs{
			NoQuant: true,
		},
	).Result()

	if err != nil {
		panic(err)
	}

	fmt.Println(res2) // >>> true

	embNoQ, err := rdb.VEmb(ctx, "quantSetNoQ", "quantElement", false).Result()

	if err != nil {
		panic(err)
	}

	fmt.Printf("NOQUANT embedding: %v\n", embNoQ)
	// >>> NOQUANT embedding: [1.262185 1.958231]

	// Add with BIN quantization
	res3, err := rdb.VAddWithArgs(ctx, "quantSetBin", "quantElement", vecQ,
		&redis.VAddArgs{
			Bin: true,
		},
	).Result()

	if err != nil {
		panic(err)
	}

	fmt.Println(res3) // >>> true

	embBin, err := rdb.VEmb(ctx, "quantSetBin", "quantElement", false).Result()

	if err != nil {
		panic(err)
	}

	fmt.Printf("BIN embedding: %v\n", embBin)
	// >>> BIN embedding: [1 1]

}

func ExampleClient_vectorset_dimension_reduction() {
	ctx := context.Background()

	rdb := redis.NewClient(&redis.Options{
		Addr:     "localhost:6379",
		Password: "", // no password set
		DB:       0,  // use default DB
	})

	defer rdb.Close()

	// Create a vector with 300 dimensions
	values := make([]float64, 300)

	for i := 0; i < 300; i++ {
		values[i] = float64(i) / 299
	}

	vecLarge := &redis.VectorValues{Val: values}

	// Add without reduction
	res1, err := rdb.VAdd(ctx, "setNotReduced", "element", vecLarge).Result()

	if err != nil {
		panic(err)
	}

	fmt.Println(res1) // >>> true

	dim1, err := rdb.VDim(ctx, "setNotReduced").Result()

	if err != nil {
		panic(err)
	}

	fmt.Printf("Dimension without reduction: %d\n", dim1)
	// >>> Dimension without reduction: 300

	// Add with reduction to 100 dimensions
	res2, err := rdb.VAddWithArgs(ctx, "setReduced", "element", vecLarge,
		&redis.VAddArgs{
			Reduce: 100,
		},
	).Result()

	if err != nil {
		panic(err)
	}

	fmt.Println(res2) // >>> true

	dim2, err := rdb.VDim(ctx, "setReduced").Result()

	if err != nil {
		panic(err)
	}

	fmt.Printf("Dimension after reduction: %d\n", dim2)
	// >>> Dimension after reduction: 100

}

Deletion performance

Deleting large vector sets using the DEL can cause latency spikes:

Redis must unlink and restructure many graph nodes.
Latency is most noticeable when deleting millions of elements.

Save and load performance

Vector sets save and load the full HNSW graph structure:

When reloading from disk is fast and there's no need to rebuild the graph.

Example: A 3M vector set with 300 components loads in ~15 seconds.

Summary of tuning tips

Factor	Effect on performance	Tip
`EF`	Slower queries but higher recall	Start low (for example, 200) and tune upward
`M`	More memory per node, better recall	Use defaults unless recall is too low
Quant type	Binary is fastest, `FP32` is slowest	Use `Q8` or `BIN` unless full precision needed
`CAS`	Faster insertions with threading	Use when high write throughput is needed

Products

Tools

Get Redis

Connect

Learn

Latest

See how it works