Understanding Pinecone cost

For the latest pricing details, see Pricing.

Pinecone serverless is usage-based, so you pay only for the data you store and the operations you run. Idle indexes cost nothing. Most early and small workloads fit within the free Starter plan, and you can lower costs further as you scale.

• Start free. The Starter plan has no monthly minimum, so you can build and test before committing to any spend.
• Prepaid credits and annual commitments. Committing usage upfront earns discounted rates. See Prepaid credits.
• Optimize your workload. Use namespaces and right-size your reads to cut ongoing query cost. See Save on costs.
• Talk to us. Standard and Enterprise customers can contact Support to optimize costs and discuss volume discounts.

The Builder, Standard, and Enterprise pricing plans include a monthly minimum usage commitment:

On the Builder plan, the monthly minimum is a flat fee that covers included usage; additional usage beyond Builder limits is blocked rather than billed. On the Standard and Enterprise plans, customers are charged for what they use each month beyond the monthly minimum.

The minimum is a commitment you grow into rather than an extra charge. Once your usage exceeds the minimum, you pay only for what you use.

Examples

Pinecone offers an incentive for customers who purchase prepaid credits with an upfront payment. Customers may purchase between $8,000 and $25,000 in prepaid credits.

Customers who purchase prepaid credits can unlock additional usage capacity at no extra cost. The available benefits vary based on the selected plan and prepaid amount.

Prepaid credits apply to Pinecone services at List Price. Any usage that exceeds the available prepaid credits will be billed at full List Price.

Customers on Standard and Enterprise pay-as-you-go plans can purchase prepaid credits directly by navigating in the Pinecone console to Settings > Billing > Plans.

With serverless indexes, you pay for the amount of data stored and operations performed, based on four usage metrics: read units, write units, storage, and egress.

Full-text search on a document index, semantic search on a vector index, and hybrid search that combines the two are all metered by the same four metrics.

For the latest serverless pricing rates, see Pricing.

A read unit (RU) is the unit Pinecone uses to measure and price the cost of a read request. Read units (RUs) measure the compute, I/O, and network resources consumed by the following read requests:

• Query
• Fetch
• List
• Full-text search

The cost of a query scales linearly with the size of the targeted namespace. Specifically, a query uses 1 RU for every 1 GB of namespace size, with a minimum of 0.25 RUs per query.

To learn how to calculate your namespace size, see Storage.

A fetch request uses 1 RU for every 10 records fetched, for example:

Specifying a non-existent ID or adding the same ID more than once doesn't increase the number of RUs used. However, a fetch request will always use at least 1 RU.

List has a fixed cost of 1 RU per call, with up to 100 records per call.

Searching a document index uses 1 RU for every 1 GB of namespace size, with a minimum of 0.25 RUs per search, the same as a query. This holds whether the request ranks by keyword relevance or by vector similarity.

A document index can declare string fields with full_text_search enabled and a dense_vector field in one schema, and each request ranks by one of them. So a hybrid search that narrows candidates with a text-match filter and then ranks what remains with a dense_vector search is billed as one read request. Running a separate keyword search and dense search and merging the results client-side is billed as two.

A write unit (WU) is the unit Pinecone uses to measure and price the cost of a write request. Write units (WUs) measure the storage and compute resources used by the following write requests:

• Upsert
• Update
• Delete

An upsert request uses 1 WU for each 1 KB of the request, with a minimum of 5 WUs per request. When an upsert modifies an existing record, the request uses 1 WU for each 1 KB of the existing record as well.

For example, the following table shows the WUs used by upsert requests at different batch sizes and record sizes, assuming all records are new:

An update request uses 1 WU for each 1 KB of the new and existing record, with a minimum of 5 WUs per request.

For example, the following table shows the WUs used by an update at different record sizes:

A delete request uses 1 WU for each 1 KB of records deleted, with a minimum of 5 WUs per request.

For example, the following table shows the WUs used by delete requests at different batch sizes and record sizes:

Specifying a non-existent ID or adding the same ID more than once doesn't increase WU use.

Deleting a namespace or deleting all records in a namespace using deleteAll uses 5 WUs.

Storage costs are based on the size of an index on a per-gigabyte (GB) monthly rate. The size of an index is defined as the total size of its records across all namespaces. For the latest storage pricing rates, see Pricing.

A record can include a dense vector, a sparse vector, or both. Use the formula that matches your data to calculate total size:

Index size = Number of records × (
               ID size + 
               Metadata size +
               Dense vector dimensions × 4 bytes
             )

Index size = Number of records × (
               ID size + 
               Metadata size +
               Number of non-zero sparse values × 8 bytes
             )

Index size = Number of records × (
               ID size + 
               Metadata size +
               Dense vector dimensions × 4 bytes + 
               Number of non-zero sparse values × 8 bytes
             )

Egress measures the data Pinecone returns to you on serverless reads. Egress is measured in GB of total response bytes returned by an in-scope read request, proportional to the record data returned (IDs, scores, values, and metadata).

Egress is metered on read requests that return per-record data:

• Query
• Fetch (by ID and by metadata)
• List
• Text search on integrated indexes
• Full-text search

Write requests (upsert, update, delete, import), index statistics (such as describe_index_stats), and index management requests aren't metered for egress.

Egress applies to indexes that use dedicated read nodes as well as on-demand indexes, because it depends on the data returned to you rather than on the hardware serving the read.

Each plan includes a monthly egress allowance, which resets at the start of each billing period:

What happens past the allowance depends on your plan:

• Usage-based plans (Standard, Enterprise): egress beyond the allowance is billed at the per-GB overage rate and reads keep serving. For the latest egress rate, see Pricing.
• Flat-fee plans (Starter, Builder): in-scope reads are blocked with a RESOURCE_EXHAUSTED (429) error and an upgrade prompt once the allowance is reached. Index statistics, index management, and write requests remain available, and the allowance resets at the start of the next billing period.

Importing from object storage is the most efficient and cost-effective method to load large numbers of records into an index. The cost of an import is based on the size of the records read, whether the records were imported successfully or not.

If the import operation fails (e.g., after encountering a vector of the wrong dimension in an import with on_error="abort"), you will still be charged for the records read. However, if the import fails because of an internal system error, you won't incur charges. In this case, the import will return the error message "We were unable to process your request. If the problem persists, please contact us at https://support.pinecone.io".

For the latest import pricing rates, see Pricing.

A backup is a static copy of a serverless index. Both the cost of storing a backup and restoring an index from a backup is based on the size of the index. For the latest backup and restore pricing rates, see Pricing.

Pinecone hosts several embedding models so it's easy to manage your vector storage and search process on a single platform. You can use a hosted model to embed your data as an integrated part of upserting and querying, or you can use a hosted model to embed your data as a standalone operation.

Embedding costs are determined by how many tokens are in a request. In general, the more words contained in your passage or query, the more tokens you generate.

For example, if you generate embeddings for the query, "What is the maximum diameter of a red pine?", Pinecone Inference generates 10 tokens, then converts them into an embedding. If the price per token for your billing plan is $.08 per million tokens, then this API call costs $.00001.

To learn more about tokenization, see Choosing an embedding model. For the latest embed pricing rates, see Pricing.

Pinecone hosts several reranking models so it's easy to manage two-stage vector retrieval on a single platform. You can use a hosted model to rerank results as an integrated part of a query, or you can use a hosted model to rerank results as a standalone operation.

Reranking costs are determined by the number of requests to the reranking model. For the latest rerank pricing rates, see Pricing.

For details on how costs are incurred in Pinecone Assistant, see Assistant pricing.

Full HIPAA compliance is included with the Enterprise plan.

For Standard plan customers, HIPAA compliance is available as an optional add-on for $190 per month. The add-on is billed monthly and added to your regular invoice. A 6-month minimum period is required.

The HIPAA compliance add-on includes:

• HIPAA-ready infrastructure
• Encrypted data storage
• Audit logging
• Enhanced security controls
• BAA execution and compliance documentation support

To enable the HIPAA compliance add-on, submit a HIPAA request. The Pinecone team will review your request and guide you through activation.

• Manage cost
• Monitor usage
• Full-text search
• Hybrid search
• Pricing