Pinecone Database limits

This page describes different types of limits for Pinecone Database.

Looking for a specific limit?

To compare monthly included usage by plan, start with read units, write units, and model usage limits.
If you received a 429 error, check rate limits, especially request-per-second limits for query, upsert, update, delete, fetch, and list.
For projects, users, indexes, namespaces, storage, backups, and collections, see object limits.
For batch sizes, metadata filters, and identifier lengths, see operation limits and identifier limits.

Rate limits

Rate limits help protect your applications from misuse and maintain the health of our shared serverless infrastructure. These limits are designed to support typical production workloads while ensuring reliable performance for all users. Most rate limits can be adjusted upon request. If you need higher limits to scale your application, contact Support with details about your use case. Pinecone is committed to supporting your growth and can often accommodate higher throughput requirements. Rate limits vary based on pricing plan and apply to serverless indexes only.

Indexes built on Dedicated Read Nodes are not subject to read unit limits for query, fetch, and list operations. For sizing and capacity planning guidance, see the Dedicated Read Nodes guide.

Data plane operations: request-per-second limits

Pinecone enforces rate limits on the number of API requests per second at the namespace level for data plane operations (query, upsert, delete, and update). These limits provide protection against excessive request rates.

Affected operations

The following operations are subject to request-per-second rate limiting:

Operation	Scope	Limit
Query	Per namespace	100
Upsert	Per namespace	100
Delete	Per namespace	100
Update	Per namespace	100

Error response

When you exceed the request-per-second limit, you’ll receive an HTTP 429 - TOO_MANY_REQUESTS response. The error message indicates which operation exceeded the limit and includes the namespace name and limit value. See the individual limit sections below for specific error message formats.

How request-per-second limits work with limits on read and write units

Request-per-second limits are enforced in addition to existing read unit and write unit limits. Requests must not exceed any applicable limits:

Index-level limits - read and write unit limits, per index
Namespace-level limits - read and write unit limits, per namespace
Request-per-second limits - requests per second, per namespace

If any limit is exceeded, the request fails with a 429 error.

Recommendations

If you’re hitting request-per-second limits:

Implement retry logic. Use exponential backoff to handle rate limit errors gracefully. See Error Handling Guide.
Pace your requests. Add client-side rate limiting to stay under limits.
Consider Dedicated Read Nodes, which don’t have request-per-second limits and provide dedicated capacity for high-throughput workloads.
If you need higher limits, contact Support with details about your use case.

All rate limits

Monthly usage limits

Metric	Starter plan	Builder plan	Standard plan	Enterprise plan
Read units per month per org	1,000,000	2,000,000	Unlimited	Unlimited
Write units per month per org	2,000,000	5,000,000	Unlimited	Unlimited
Embedding tokens per month per model	5,000,000	10,000,000	Unlimited	Unlimited
Rerank requests per month per model	Model-specific	Model-specific	Model-specific	Model-specific

Data operation throughput limits

Metric	Starter plan	Builder plan	Standard plan	Enterprise plan
Upsert size per second per namespace	50 MB	50 MB	50 MB	50 MB
Query read units per second per index	2,000	2,000	2,000	2,000
Query requests per second per namespace	100	100	100	100
Update records per second per namespace	100	100	100	100
Update requests per second per namespace	100	100	100	100
Update by metadata requests per second per namespace	5	5	5	5
Update by metadata requests per second per index	500	500	500	500
Upsert requests per second per namespace	100	100	100	100
Fetch requests per second per index	100	100	100	100
List requests per second per index	200	200	200	200
Describe index stats requests per second per index	100	100	100	100
Delete requests per second per namespace	100	100	100	100
Delete records per second per namespace	5,000	5,000	5,000	5,000
Delete records per second per index	5,000	5,000	5,000	5,000
Delete by metadata requests per second per namespace	5	5	5	5
Delete by metadata requests per second per index	500	500	500	500

Model throughput limits

Metric	Starter plan	Builder plan	Standard plan	Enterprise plan
Embedding tokens per minute per model	Model-specific	Model-specific	Model-specific	Model-specific
Rerank requests per minute per model	Model-specific	Model-specific	Model-specific	Model-specific

Read units per month per org

Starter plan	Builder plan	Standard plan	Enterprise plan
1,000,000	2,000,000	Unlimited	Unlimited

Read units measure the compute, I/O, and network resources used by fetch, query, and list requests to serverless indexes. When you reach the monthly read unit limit for an organization, fetch, query, and list requests to serverless indexes in the organization will fail and return a 429 - TOO_MANY_REQUESTS status with the following error:

Error response

Request failed. You've reached your read unit limit for the current month limit. 
To continue reading data, upgrade your plan. 

To continue reading from serverless indexes in the organization, upgrade your plan. To check how close you are to the monthly read unit limit for your organization, do the following:

Open the Pinecone console.
Select the project.
Select any index in the project.
Look under Usage.

Indexes built on Dedicated Read Nodes are not subject to read unit limits for query, fetch, and list operations. For sizing and capacity planning guidance, see the Dedicated Read Nodes guide.

Write units per month per org

Starter plan	Builder plan	Standard plan	Enterprise plan
2,000,000	5,000,000	Unlimited	Unlimited

Write units measure the storage and compute resources used by upsert, update, and delete requests to serverless indexes. When you reach the monthly write unit limit for an organization, upsert, update, and delete requests to serverless indexes in the organization will fail and return a 429 - TOO_MANY_REQUESTS status with the following error:

Error response

Request failed. You've reached your write unit limit for the current month. 
To continue writing data, upgrade your plan.

To continue writing data to serverless indexes in the organization, upgrade your plan. To check how close you are to the monthly write unit limit for your organization, do the following:

Open the Pinecone console.
Select the project.
Select any index in the project.
Look under Usage.

Upsert size per second per namespace

Starter plan	Builder plan	Standard plan	Enterprise plan
50 MB	50 MB	50 MB	50 MB

When you reach the per second upsert size for a namespace in an index, additional upserts will fail and return a 429 - TOO_MANY_REQUESTS status with the following error:

Error response

Request failed. You've reached the max upsert size limit per second for index <index name>. 
Pace your upserts or contact Pinecone Support (https://app.pinecone.io/organizations/-/settings/support/ticket) to request a higher limit.

To handle this limit, implement retry logic with exponential backoff. If you need a higher limit for your use case, contact Support.

Query read units per second per index

Starter plan	Builder plan	Standard plan	Enterprise plan
2,000	2,000	2,000	2,000

Pinecone measures query usage in read units. When you reach the per second limit for queries across all namespaces in an index, additional queries will fail and return a 429 - TOO_MANY_REQUESTS status with the following error:

Error response

Request failed. You've reached the max query read units per second for index <index name>. 
Pace your queries or contact Pinecone Support (https://app.pinecone.io/organizations/-/settings/support/ticket) to request a higher limit.

To handle this limit, implement retry logic with exponential backoff. If you need a higher limit for your use case, contact Support. To check how many read units a query consumes, check the query response.

Indexes built on Dedicated Read Nodes are not subject to read unit limits for query, fetch, and list operations. For sizing and capacity planning guidance, see the Dedicated Read Nodes guide.

Query requests per second per namespace

Starter plan	Builder plan	Standard plan	Enterprise plan
100	100	100	100

When you reach the per second query limit for a namespace in an index, additional queries will fail and return a 429 - TOO_MANY_REQUESTS status with the following error:

Error response

Request failed. You've reached the query QPS limit for namespace {namespace_name} ({limit} QPS). Pace your queries,
consider Dedicated Read Nodes for your index, or contact Pinecone Support
(https://app.pinecone.io/organizations/-/settings/support/ticket) to request a higher limit.

To handle this limit, implement retry logic with exponential backoff. If you need a higher limit for your use case, contact Support.

Indexes built on Dedicated Read Nodes are not subject to read unit limits for query, fetch, and list operations. For sizing and capacity planning guidance, see the Dedicated Read Nodes guide.

Update records per second per namespace

Starter plan	Builder plan	Standard plan	Enterprise plan
100	100	100	100

When you reach the per second update limit for a namespace in an index, additional updates will fail and return a 429 - TOO_MANY_REQUESTS status with the following error:

Error response

Request failed. You've reached the max update records per second for namespace <namespace name>. 
Pace your update requests or contact Pinecone Support (https://app.pinecone.io/organizations/-/settings/support/ticket) to request a higher limit.

To handle this limit, implement retry logic with exponential backoff. If you need a higher limit for your use case, contact Support.

Update requests per second per namespace

Starter plan	Builder plan	Standard plan	Enterprise plan
100	100	100	100

When you reach the per second update request limit for a namespace in an index, additional update requests will fail and return a 429 - TOO_MANY_REQUESTS status with the following error:

Error response

Request failed. You've reached the update QPS limit for namespace {namespace_name} ({limit} QPS). Pace your update requests or contact Pinecone Support (https://app.pinecone.io/organizations/-/settings/support/ticket) to request a higher limit.

To handle this limit, implement retry logic with exponential backoff. If you need a higher limit for your use case, contact Support.

If you need to update most of the records in a large namespace, contact Support for help creating an export to enable a faster and more cost-effective approach.

Update by metadata requests per second per namespace

Starter plan	Builder plan	Standard plan	Enterprise plan
5	5	5	5

When you reach the per second update by metadata request limit for a namespace in an index, additional update by metadata requests will fail and return a 429 - TOO_MANY_REQUESTS status with the following error:

Error response

Request failed. You've reached the max update by metadata requests per second for namespace <namespace name>. Pace your update by metadata requests or contact Pinecone Support (https://app.pinecone.io/organizations/-/settings/support/ticket) to request a higher limit.

To handle this limit, implement retry logic with exponential backoff. If you need a higher limit for your use case, contact Support.

Update by metadata requests per second per index

Starter plan	Builder plan	Standard plan	Enterprise plan
500	500	500	500

When you reach the per second update by metadata request limit across all namespaces in an index, additional update by metadata requests will fail and return a 429 - TOO_MANY_REQUESTS status with the following error:

Error response

Request failed. You've reached the max update by metadata requests per second for index <index name>. Pace your update by metadata requests or contact Pinecone Support (https://app.pinecone.io/organizations/-/settings/support/ticket) to request a higher limit.

To handle this limit, implement retry logic with exponential backoff. If you need a higher limit for your use case, contact Support.

Upsert requests per second per namespace

Starter plan	Builder plan	Standard plan	Enterprise plan
100	100	100	100

When you reach the per second upsert request limit for a namespace in an index, additional upsert requests will fail and return a 429 - TOO_MANY_REQUESTS status with the following error:

Error response

Request failed. You've reached the upsert QPS limit for namespace {namespace_name} ({limit} QPS). Pace your upsert requests or contact Pinecone Support (https://app.pinecone.io/organizations/-/settings/support/ticket) to request a higher limit.

To handle this limit, implement retry logic with exponential backoff. If you need a higher limit for your use case, contact Support.

Fetch requests per second per index

Starter plan	Builder plan	Standard plan	Enterprise plan
100	100	100	100

When you reach the per second fetch limit across all namespaces in an index, additional fetch requests will fail and return a 429 - TOO_MANY_REQUESTS status with the following error:

Error response

Request failed. You've reached the max fetch requests per second for index <index name>.
Pace your fetch requests or contact Pinecone Support (https://app.pinecone.io/organizations/-/settings/support/ticket) to request a higher limit.

To handle this limit, implement retry logic with exponential backoff. If you need a higher limit for your use case, contact Support.

Indexes built on Dedicated Read Nodes are not subject to read unit limits for query, fetch, and list operations. For sizing and capacity planning guidance, see the Dedicated Read Nodes guide.

List requests per second per index

Starter plan	Builder plan	Standard plan	Enterprise plan
200	200	200	200

When you reach the per second list limit across all namespaces in an index, additional list requests will fail and return a 429 - TOO_MANY_REQUESTS status with the following error:

Error response

Request failed. You've reached the max list requests per second for index <index name>.
Pace your list requests or contact Pinecone Support (https://app.pinecone.io/organizations/-/settings/support/ticket) to request a higher limit.

To handle this limit, implement retry logic with exponential backoff. If you need a higher limit for your use case, contact Support.

Indexes built on Dedicated Read Nodes are not subject to read unit limits for query, fetch, and list operations. For sizing and capacity planning guidance, see the Dedicated Read Nodes guide.

Describe index stats requests per second per index

Starter plan	Builder plan	Standard plan	Enterprise plan
100	100	100	100

When you reach the per second describe index stats limit across all namespaces in an index, additional describe index stats requests will fail and return a 429 - TOO_MANY_REQUESTS status with the following error:

Error response

Request failed. You've reached the max describe_index_stats requests per second for index <index>. 
Pace your describe_index_stats requests or contact Pinecone Support (https://app.pinecone.io/organizations/-/settings/support/ticket) to request a higher limit.

To handle this limit, implement retry logic with exponential backoff. If you need a higher limit for your use case, contact Support.

Delete requests per second per namespace

Starter plan	Builder plan	Standard plan	Enterprise plan
100	100	100	100

When you reach the per second delete request limit for a namespace in an index, additional delete requests will fail and return a 429 - TOO_MANY_REQUESTS status with the following error:

Error response

Request failed. You've reached the delete QPS limit for namespace {namespace_name} ({limit} QPS). Pace your delete requests or contact Pinecone Support (https://app.pinecone.io/organizations/-/settings/support/ticket) to request a higher limit.

To handle this limit, implement retry logic with exponential backoff. If you need a higher limit for your use case, contact Support.

Delete records per second per namespace

Starter plan	Builder plan	Standard plan	Enterprise plan
5000	5000	5000	5000

When you reach the per second delete limit for a namespace in an index, additional deletes will fail and return a 429 - TOO_MANY_REQUESTS status with the following error:

Error response

Request failed. You've reached the max delete records per second for namespace <namespace name>. 
Pace your delete requests or contact Pinecone Support (https://app.pinecone.io/organizations/-/settings/support/ticket) to request a higher limit.

To handle this limit, implement retry logic with exponential backoff. If you need a higher limit for your use case, contact Support.

Delete records per second per index

Starter plan	Builder plan	Standard plan	Enterprise plan
5000	5000	5000	5000

When you reach the per second delete limit across all namespaces in an index, additional deletes will fail and return a 429 - TOO_MANY_REQUESTS status with the following error:

Error response

Request failed. You've reached the max delete records per second for index <index name>. 
Pace your delete requests or contact Pinecone Support (https://app.pinecone.io/organizations/-/settings/support/ticket) to request a higher limit.

To handle this limit, implement retry logic with exponential backoff. If you need a higher limit for your use case, contact Support.

Delete by metadata requests per second per namespace

Starter plan	Builder plan	Standard plan	Enterprise plan
5	5	5	5

When you reach the per second delete by metadata request limit for a namespace in an index, additional delete by metadata requests will fail and return a 429 - TOO_MANY_REQUESTS status with the following error:

Error response

Request failed. You've reached the max delete by metadata requests per second for namespace <namespace name>. Pace your delete by metadata requests or contact Pinecone Support (https://app.pinecone.io/organizations/-/settings/support/ticket) to request a higher limit.

To handle this limit, implement retry logic with exponential backoff. If you need a higher limit for your use case, contact Support.

Delete by metadata requests per second per index

Starter plan	Builder plan	Standard plan	Enterprise plan
500	500	500	500

When you reach the per second delete by metadata request limit across all namespaces in an index, additional delete by metadata requests will fail and return a 429 - TOO_MANY_REQUESTS status with the following error:

Error response

Request failed. You've reached the max delete by metadata requests per second for index <index name>. Pace your delete by metadata requests or contact Pinecone Support (https://app.pinecone.io/organizations/-/settings/support/ticket) to request a higher limit.

To handle this limit, implement retry logic with exponential backoff. If you need a higher limit for your use case, contact Support.

Embedding tokens per minute per model

Embedding model	Input type	Starter plan	Builder plan	Standard plan	Enterprise plan
`llama-text-embed-v2`	Passage	250,000	250,000	1,000,000	1,000,000
	Query	50,000	50,000	250,000	250,000
`multilingual-e5-large`	Passage	250,000	250,000	1,000,000	1,000,000
	Query	50,000	50,000	250,000	250,000
`pinecone-sparse-english-v0`	Passage	250,000	250,000	3,000,000	3,000,000
	Query	250,000	250,000	3,000,000	3,000,000

When you reach the per minute token limit for an embedding model hosted by Pinecone, additional embeddings will fail and return a 429 - TOO_MANY_REQUESTS status with the following error:

Error response

Request failed. You've reached the max embedding tokens per minute (<limit>) model '<model name>'' and input type '<passage|query>' for the current project. 
To increase this limit, upgrade your plan.

To increase this limit, upgrade your plan. Otherwise, you can handle this limit by implementing retry logic with exponential backoff.

Embedding tokens per month per model

Starter plan	Builder plan	Standard plan	Enterprise plan
5,000,000	10,000,000	Unlimited	Unlimited

When you reach the monthly token limit for an embedding model hosted by Pinecone, additional embeddings will fail and return a 429 - TOO_MANY_REQUESTS status with the following error:

Error response

Request failed. You've reached the embedding token limit (<limit>) for model <model name> for the current month. 
To continue using this model, upgrade your plan.

To increase this limit, upgrade your plan or contact Support.

Rerank requests per minute per model

Reranking model	Starter plan	Builder plan	Standard plan	Enterprise plan
`cohere-rerank-4-fast`	Not available	Not available	300	300
`cohere-rerank-3.5`	Not available	Not available	300	300
`bge-reranker-v2-m3`	60	60	60	60
`pinecone-rerank-v0`	60	Not available	60	60

When you reach the per minute request limit for a reranking model hosted by Pinecone, additional reranking requests will fail and return a 429 - TOO_MANY_REQUESTS status with the following error:

Error response

Request failed. You've reached the max rerank requests per minute (<limit>) for model '<model name>' for the current project. 
To increase this limit, upgrade your plan.

To increase this limit, upgrade your plan.

Rerank requests per month per model

Reranking model	Starter plan	Builder plan	Standard plan	Enterprise plan
`cohere-rerank-4-fast`	Not available	Not available	Unlimited	Unlimited
`cohere-rerank-3.5`	Not available	Not available	Unlimited	Unlimited
`bge-reranker-v2-m3`	500	1,000	Unlimited	Unlimited
`pinecone-rerank-v0`	500	Not available	Unlimited	Unlimited

When you reach the monthly request limit for a reranking model hosted by Pinecone, additional reranking requests will fail and return a 429 - TOO_MANY_REQUESTS status with the following error:

Error response

Request failed. You've reached the rerank request limit (<limit>) for model <model name> for the current month. 
To continue using this model, upgrade your plan.

To increase this limit, upgrade your plan or contact Support.

Inference requests per second or minute, per project

Metric	Starter plan	Builder plan	Standard plan	Enterprise plan
Inference requests per second	100	100	100	100
Inference requests per minute	2000	2000	2000	2000

When you reach the per second or per minute request limit, inference requests will fail and return a 429 - TOO_MANY_REQUESTS status with the following error:

Error response

Request failed. You've reached the max inference requests per second (<limit>) for the current project.

This error indicates per second or per minute, as applicable.

To handle this limit, implement retry logic with exponential backoff.

Object limits

Object limits are restrictions on the number or size of objects in Pinecone. Object limits vary based on pricing plan.

Metric	Starter plan	Builder plan	Standard plan	Enterprise plan
Projects per organization	1	5	20	100
Users per organization	2	5	Unlimited	Unlimited
Serverless indexes per project ¹	5	10	20	200
Serverless index storage per org	2 GB	10 GB	N/A	N/A
Namespaces per serverless index	100	1,000	100,000	100,000
Serverless backups per project	N/A	N/A	500	1000
Collections per project	100	N/A	N/A	N/A

^{1 On the Starter plan, all serverless indexes must be in the us-east-1 region of AWS. Builder, Standard, and Enterprise plans can create indexes in any supported region.}

Projects per organization

Starter plan	Builder plan	Standard plan	Enterprise plan
1	5	20	100

When you reach this quota for an organization, trying to create projects will fail and return a 403 - QUOTA_EXCEEDED status with the following error:

Error response

Request failed. You've reached the max projects allowed in organization <org name>. 
To add more projects, upgrade your plan.

To increase this quota, upgrade your plan or contact Support.

Users per organization

Starter plan	Builder plan	Standard plan	Enterprise plan
2	5	Unlimited	Unlimited

When you reach this quota for an organization, trying to add users to the organization will fail. To increase this quota, upgrade your plan.

Serverless indexes per project

Starter plan	Builder plan	Standard plan	Enterprise plan
5	10	20	200

When you reach this quota for a project, trying to create serverless indexes in the project will fail and return a 403 - QUOTA_EXCEEDED status with the following error:

Error response

Request failed. You've reached the max serverless indexes allowed in project <project>. 
Use namespaces to partition your data into logical groups, or upgrade your plan to add more serverless indexes.

To stay under this quota, consider using namespaces instead of creating multiple indexes. Namespaces let you partition your data into logical groups within a single index. This approach not only helps you stay within index limits, but can also improve query performance and lower costs by limiting searches to relevant data subsets. To increase this quota, upgrade your plan.

Serverless index storage per org

This limit applies to organizations on the Starter and Builder plans only.

Starter plan	Builder plan	Standard plan	Enterprise plan
2 GB	10 GB	N/A	N/A

When you’ve reached this quota for an organization, updates and upserts into serverless indexes will fail and return a 403 - QUOTA_EXCEEDED status with the following error:

Error response

Request failed. You've reached the max storage allowed for organization <org name>. 
To update or upsert new data, delete records or upgrade your plan.

To continue writing data into your serverless indexes, delete records to bring your organization under the limit or upgrade your plan.

Namespaces per serverless index

Starter plan	Builder plan	Standard plan	Enterprise plan
100	1,000	100,000	100,000

When you reach this quota for a serverless index, trying to upsert records into a new namespace in the index will fail and return a 403 - QUOTA_EXCEEDED status with the following error:

Error response

Request failed. You've reached the max namespaces allowed in serverless index <index name>. 
To add more namespaces, upgrade your plan.

To increase this quota, upgrade your plan.

Namespaces per serverless index vary by plan. On the Standard and Enterprise plans, Pinecone can accommodate million-scale namespaces and beyond for specific use cases. If your application requires more than 100,000 namespaces, contact Support.

Serverless backups per project

Starter plan	Builder plan	Standard plan	Enterprise plan
N/A	N/A	500	1000

Backups are not available on the Starter or Builder plans. On the Standard and Enterprise plans, when you reach this quota for a project, trying to create serverless backups in the project will fail and return a 403 - QUOTA_EXCEEDED status with the following error:

Error response

Backup failed to create. Quota for number of backups per index exceeded.

Collections per project

Starter plan	Builder plan	Standard plan	Enterprise plan
100	N/A	N/A	N/A

When you reach this quota for a project, trying to create collections in the project will fail and return a 403 - QUOTA_EXCEEDED status with the following error:

Error response

Request failed. You've reached the max collections allowed in project <project name>. 
To add more collections, upgrade your plan.

To increase this quota, upgrade your plan.

Operation limits

Operation limits are restrictions on the size, number, or other characteristics of operations in Pinecone. Operation limits are fixed and do not vary based on pricing plan.

Upsert limits

Metric	Limit
Max batch size	2 MB or 1000 records with vectors 96 records with text
Max documents per upsert request	1000
Max document upsert request size	2 MB
Max document size	2 MB
Max `full_text_search` string fields per schema	100
Max size per `full_text_search` string field	100 KB
Max tokens per `full_text_search` string field	10,000
Max bytes per token	256 bytes
Max filterable metadata size per document	40 KB
Max length for a record ID	512 characters
Max dimensionality for dense vectors	20,000
Max non-zero values for sparse vectors	2048
Max dimensionality for sparse vectors	4.2 billion

The 40 KB filterable metadata limit does not apply to full_text_search text fields.

Import limits

If your import exceeds these limits, you’ll get an error specifying the limit exceeded. See Troubleshooting for details.

Metric	Limit
Max namespaces per import	10,000
Max total input data size (on-demand indexes)	1 TB
Max total input data size (DRN indexes)	Unlimited
Max files per import	100,000
Max size per file	10 GB

The total input data size limit does not apply to indexes with dedicated read nodes. Bulk import supports indexes without a schema definition (Parquet files) and indexes with document schemas (JSONL files). Semantic-text (auto-embedded) fields are not yet supported in document schemas.

Query limits

Metric	Limit
Max `top_k` value	10,000
Max result size	4MB

The query result size is affected by the dimension of the dense vectors and whether or not dense vector values and metadata are included in the result.

If a query fails due to exceeding the 4MB result size limit, choose a lower top_k value, or use include_metadata=False or include_values=False to exclude metadata or values from the result. For better performance, especially with higher top_k values, avoid including vector values unless you need them.

Fetch limits

Fetch by ID limits:

Metric	Limit
Max record IDs per fetch request	1,000

Fetch by metadata limits:

Metric	Limit
Max records per response	10,000
Max response size	4 MB
Max request rate	5 requests per second per namespace

To retrieve more than 10,000 matching records, paginate through results using the paginationToken parameter. See Fetch records by metadata.

Delete limits

Metric	Limit
Max record IDs per delete request	1,000

Metadata filter limits

The following limits apply to metadata filter expressions used in query, delete, update, and fetch operations.

Limit	Value	Description
Maximum values per `$in` or `$nin` operator	10,000	Each `$in` or `$nin` operator accepts up to 10,000 values in its array. This limit applies per operator—if you have multiple `$in` operators in a single filter, each is independently limited to 10,000 values.

When you exceed this limit, the request will fail and return a 400 - BAD_REQUEST error.

Rationale

Large $in operators can impact query performance and cost. Filters with thousands of values increase request payload size and end-to-end latency. Additionally, using large filters typically indicates a shared namespace architecture, which increases query costs—queries scan the entire namespace regardless of filters.

Alternative approaches

If you need to filter by more than 10,000 values, consider these alternatives:

Use namespaces for tenant isolation: Instead of filtering by tenant IDs within a single namespace, create separate namespaces for each tenant or tenant group. This can also reduce query costs. See Design for multi-tenancy.
Use broader access control groups: Instead of filtering by individual user IDs, filter by organization, project, or role. This reduces the number of values in your $in filter. See Design for multi-tenancy.
Post-filter client-side: Retrieve a larger top K without filtering (for example, top 1000), then filter results client-side.
Run multiple queries: Split your filter into multiple queries with smaller $in operators and combine the results client-side.

To avoid hitting this limit in production, validate the size of your $in and $nin arrays in your application code before making the request to Pinecone.

Identifier limits

An identifier is a string of characters used to identify “named” objects in Pinecone. The following Pinecone objects use strings as identifiers:

Object	Field	Max # characters	Allowed characters
Organization	`name`	512	UTF-8 except `\0` Cannot be empty
Project	`name`	512	UTF-8 except `\0` Cannot be empty
Index	`name`	45	`a-z`, `0-9`, and `-` Must be lowercase Cannot start or end with `-` Cannot be empty
Namespace	`namespace`	512	ASCII except `\0` For the default namespace, use `""` (or `"__default__"`, in API versions `2025-04` and later)
Record	`id`	512	ASCII except `\0` Cannot be empty

​Rate limits

​Data plane operations: request-per-second limits

​Affected operations

​Error response

​How request-per-second limits work with limits on read and write units

​Recommendations

​All rate limits

​Monthly usage limits

​Data operation throughput limits

​Model throughput limits

​Read units per month per org

​Write units per month per org

​Upsert size per second per namespace

​Query read units per second per index

​Query requests per second per namespace

​Update records per second per namespace

​Update requests per second per namespace

​Update by metadata requests per second per namespace

​Update by metadata requests per second per index

​Upsert requests per second per namespace

​Fetch requests per second per index

​List requests per second per index

​Describe index stats requests per second per index

​Delete requests per second per namespace

​Delete records per second per namespace

​Delete records per second per index

​Delete by metadata requests per second per namespace

​Delete by metadata requests per second per index

​Embedding tokens per minute per model

​Embedding tokens per month per model

​Rerank requests per minute per model

​Rerank requests per month per model

​Inference requests per second or minute, per project

​Object limits

​Projects per organization

​Users per organization

​Serverless indexes per project

​Serverless index storage per org

​Namespaces per serverless index

​Serverless backups per project

​Collections per project

​Operation limits

​Upsert limits

​Import limits

​Query limits

​Fetch limits

​Delete limits

​Metadata filter limits

​Rationale

​Alternative approaches

​Identifier limits

Rate limits

Data plane operations: request-per-second limits

Affected operations

Error response

How request-per-second limits work with limits on read and write units

Recommendations

All rate limits

Monthly usage limits

Data operation throughput limits

Model throughput limits

Read units per month per org

Write units per month per org

Upsert size per second per namespace

Query read units per second per index

Query requests per second per namespace

Update records per second per namespace

Update requests per second per namespace

Update by metadata requests per second per namespace

Update by metadata requests per second per index

Upsert requests per second per namespace

Fetch requests per second per index

List requests per second per index

Describe index stats requests per second per index

Delete requests per second per namespace

Delete records per second per namespace

Delete records per second per index

Delete by metadata requests per second per namespace

Delete by metadata requests per second per index

Embedding tokens per minute per model

Embedding tokens per month per model

Rerank requests per minute per model

Rerank requests per month per model

Inference requests per second or minute, per project

Object limits

Projects per organization

Users per organization

Serverless indexes per project

Serverless index storage per org

Namespaces per serverless index

Serverless backups per project

Collections per project

Operation limits

Upsert limits

Import limits

Query limits

Fetch limits

Delete limits

Metadata filter limits

Rationale

Alternative approaches

Identifier limits