Meet Zilliz at AWS re:Invent 2026 in Las Vegas, Nov 30–Dec 4.
Apply Now

$200 in free credits when you sign up with a work emailHow to read this leaderboard

Dataset Note:

  1. For "1M Dataset" below, we refer to the "Cohere Medium" (Click to copy S3 URI) Dataset (1M * 768 float dense vectors).
  2. For "10M" Dataset below, we refer to the "Cohere Large" (Click to copy S3 URI) Dataset (10M * 768 float dense vectors).
Vector Search Latency and QPS at $1,000 Monthly Cost
Copy link to this section
Dataset
P99 Latency (ms)
QPS
Recall
ZillizCloud-8cu-perf
2
13,316.23
0.9383
Milvus-16c64g-sq4u-fp16-force_merge
2.3
9,575.69
0.9189
Milvus-16c64g-sq8-force_merge
2.4
5,973
0.9192
OpenSearch-16c128g-force_merge
7.2
3,055.01
0.9066
ElasticCloud-8c60g-force_merge
17.1
2,353.89
0.9143
QdrantCloud-16c64g
6.4
1,242.43
0.9474
Pinecone-p2.x8-1node
13.7
1,146.53
0.9262
OpenSearch-16c128g
13.2
950.63
0.9140
Streaming Performance
Copy link to this section
This chart benchmarks search performance on the Cohere-10M dataset under constant ingestion pressure. We measure p99 serial search latency and maximum concurrent QPS at 90% data capacity while the insertion workload remains active. These 'streaming' results are then contrasted with the 'static' performance on a fully indexed database.
Plot values
  • static
  • constant ingestion (500 rows/s)
  • constant ingestion (1000 rows/s)

ZillizCloud

8cu-perf

7,385
Recall: 0.9384
2,119
1,860

Milvus

16c64g-sq8-force_merge

2,747
Recall: 0.9204
306
156

ElasticCloud

8c60g-force_merge

1,452
Recall: 0.9042
61.67
61.82

Pinecone

p2.x8-1node

1,131
Recall: 0.9024
367.4
369.7

TurboPuffer

2026-03-31

649.9
Recall: 0.8352
536
442.6

OpenSearch

16c128g

505.7
Recall: 0.9068
161.7
149.7

QdrantCloud

16c64g

446.9
Recall: 0.9357
393.8
347.6
Performance vs. Recall
Copy link to this section
Dataset
Plot values
Filtering Performance
Copy link to this section
Dataset
Plot values

Run VDBBench on your own data

These results use public datasets. Your vectors, filters and traffic decide what you actually get—run the same open-source benchmark on them.

Test Zilliz Cloud

Benchmark Zilliz Cloud on your data

  1. Sign up for Zilliz Cloud—$200 in free credits with a work email.
  2. Create a cluster and copy its endpoint and API key.
  3. Install VDBBench with pip install vectordb-bench and start it with init_bench.
  4. Load your vectors as Parquet files on the Custom Dataset page and run the case against your cluster.

Test other databases

Benchmark the database you use today

  1. VDBBench is open source and supports Milvus, Elasticsearch, OpenSearch, pgvector, Pinecone, Qdrant, Weaviate, turbopuffer and more.
  2. Run the same case and dataset against your current database.
  3. Put the two results side by side: QPS, P99 latency, recall and cost.

Talk to us

Run it with a Zilliz engineer

Evaluating a migration from Elasticsearch, OpenSearch or pgvector, or running more than 100M vectors? A Zilliz engineer runs the benchmark on your data with you.

See these numbers on your workload

Start on Zilliz Cloud with $200 in free credits, or size a production workload with an engineer.