When you create a Pinecone Index, you choose the pod type, number of pods, and number of replicas. You will be billed hourly with monthly invoices.
|p1||Optimized for lower latency. Used by standard RAM-based index.||1 vCPU + 4GB RAM||$0.070/hr|
|s1||Optimized for storage capacity, resulting in lower overall cost. Used by hybrid index.||1 vCPU + 20GB SSD||$0.075/hr|
More pods increase the number of vectors the index can hold. The fewer the vectors per pod, the lower the latency.
More replicas increase throughput capacity (QPS) linearly. Throughput is related to latency — lower latency provides greater throughput per replica.
Choose the pod types, number of pods, and number of replicas to meet your cost and performance requirements. You can achieve low latencies and high throughput with any number of vectors.
Contact us about dedicated environments to meet your compliance and performance requirements.
Pinecone takes security seriously for all users. Read about our security practices.