01
Overview
Pinecone Full-Text Search is generally available and is included as an index type across Pinecone's current product plans. The important pricing detail is that Pinecone does not publish one universal fee called a 'full-text search query price.' Full-text indexes use the same database capacity model as vector indexes, so cost depends on storage, reads, writes, cloud/region and the deployment model you choose.
The figures below are current public list pricing as of September 12, 2026 and should be rechecked before purchase because Pinecone can change allowances and regional rates.
02
Current Pinecone plans
Starter — Free. Pinecone lists Dense, Sparse and Full-Text indexes on the Starter plan. The current pricing page shows up to 2 GB of database storage, up to 2 million write units per month and up to 1 million read units per month, plus other product-specific allowances. Starter is intended for evaluation and small applications.
Builder — $20/month flat. Builder adds higher limits, cloud/region choice, multiple projects/users and monitoring options on top of Starter capabilities.
Standard — $50/month minimum usage. The monthly minimum is applied to usage. Pinecone currently advertises pay-as-you-go Database On-Demand, Inference and Assistant usage, plus Dedicated Read Nodes, import, backup/restore and enterprise-oriented controls such as RBAC/SSO options.
Enterprise — $500/month minimum usage. Enterprise adds features such as a 99.95% uptime SLA, BYOC, private endpoints, customer-managed encryption keys, audit logs, service accounts and other controls. The minimum is applied to usage; actual spend can exceed it.
03
How full-text database usage is metered
Pinecone's GA announcement says Full-Text Search uses the same capacity model as vector indexes. For On-Demand deployments, database usage is driven by storage, read units and write units. Pinecone's pricing page currently lists paid write-unit rates that vary by cloud and region rather than a single global number.
A read unit is a metering abstraction for database reads, not a promise that every query costs exactly one RU. Query shape, result size and workload characteristics affect usage. That is why a fixed 'cost per BM25 query' would be misleading without a specific workload and region.
04
Dedicated Read Nodes
For sustained production traffic, Pinecone offers Dedicated Read Nodes (DRNs), which use provisioned read capacity instead of purely request-based On-Demand reads. Pinecone markets DRNs for predictable throughput and potentially lower cost at sustained query volume. Any cost-reduction percentages published in Pinecone case studies are workload-specific vendor results, not a guarantee for a new deployment.
05
Full-text search does not include every AI cost
Database usage is only one part of a RAG stack. If you generate dense embeddings, sparse embeddings or rerank results using Pinecone Inference or another provider, those inference costs are separate. LLM generation is also separate unless it is part of another billed Pinecone product.
A hybrid retrieval workflow can therefore include database reads plus embedding/reranking plus model-generation charges. Budget the complete request path rather than comparing only index storage.
06
What limits matter on Starter
The current Starter plan publishes database allowances including up to 2 GB storage, 2 million write units per month and 1 million read units per month. Pinecone also gives illustrative workload examples on its pricing page, but explicitly says those examples are illustrative rather than binding quotes.
Because Full-Text Search is a relatively new GA feature, do not extrapolate those semantic-search examples into guaranteed BM25 throughput. Measure your own document sizes, query patterns and result sizes.
07
How to estimate your actual bill
Start with four inputs: corpus size, monthly writes/updates, monthly searches and expected query pattern. Then decide whether On-Demand or sustained provisioned reads fit the workload. Add any inference costs for embeddings/reranking and model-provider costs for generation.
Pinecone provides a pricing calculator for workload estimates. Use it with the actual cloud/region and deployment choice because regional rates can differ.
08
Is Pinecone Full-Text Search free?
It can be used within the free Starter plan's included allowances. That does not mean unlimited Full-Text Search is free. Once a workload requires paid-plan capabilities or exceeds relevant allowances, the applicable plan and usage model determine cost.
09
Bottom line
Pinecone Full-Text Search does not introduce a simple standalone BM25 fee. It inherits Pinecone's database economics: free limited usage for Starter, a $20 Builder tier, usage-based Standard with a $50 monthly minimum, and Enterprise beginning with a $500 monthly minimum. For production RAG and agent systems, estimate the complete retrieval and inference path rather than asking for one universal full-text query price.
Sources
Primary and supporting sources
Facts were rechecked against the linked sources immediately before publication. Pricing, product availability and rollout status can change.