2024 releases

Pinecone Assistant JSON mode and EU region deployment

Pinecone Assistant can now return a JSON response.

You can now create an assistant in the eu region.

Released Spark-Pinecone connector v1.2.0

Released v1.2.0 of the Pinecone Java SDK. This version introduces support for stream upserts with structured streaming. This enhancement allows users to seamlessly stream data into Pinecone for upsert operations.

New integration with HoneyHive

Added the HoneyHive integration page.

Released Python SDK v5.4.2

Released v5.4.2 of the Pinecone Python SDK. This release adds a required keyword argument, metric, to the query_namespaces method. This change enables the SDK to merge results no matter how many results are returned.

Launch week: Pinecone Local

Pinecone now offers Pinecone Local, an in-memory database emulator available as a Docker image. You can use Pinecone Local to develop your applications locally, or to test your applications in CI/CD, without connecting to your Pinecone account, affecting production data, or incurring any usage or storage fees. Pinecone Local is in public preview.

Launch week: Enhanced security and access controls

Support for customer-managed encryption keys (CMEK) is now in public preview.

You can now change API key permissions.

Private Endpoints are now in general availability. Use Private Endpoints to connect AWS PrivateLink to Pinecone while keeping your VPC private from the public internet.

Audit logs, now in early access, provide a detailed record of user and API actions that occur within the Pinecone platform.

Launch week: pinecone-rerank-v0 and cohere-rerank-3.5 on Pinecone Inference

Released pinecone-rerank-v0, Pinecone’s state of the art reranking model that out-performs competitors on widely accepted benchmarks. This model is in public preview.

Pinecone Inference now hosts cohere-rerank-3.5, Cohere’s leading reranking model.

Launch week: Integrated Inference

You can now use integrated Pinecone Inference to store and search data without extra steps for embedding data and reranking results.

Released .NET SDK v2.1.0

Released v2.1.0 of the Pinecone .NET SDK. This version introduces the ClientOptions.IsTlsEnabled property, which must be set to false for non-secure client connections.

Improved batch deletion guidance

Improved the guidance and example code for deleting records in batches.

Launch week: Released pinecone-sparse-english-v0

Pinecone Inference now supports pinecone-sparse-english-v0, Pinecone’s sparse embedding model, which estimates the lexical importance of tokens by leveraging their context, unlike traditional retrieval models like BM25, which rely solely on term frequency. This model is in public preview.

Pinecone docs: New workflows and best practices

Added typical Pinecone Database and Pinecone Assistant workflows to the Docs landing page.

Updated various examples to use the production best practice of targeting an index by host instead of name.

Updated the Amazon Bedrock integration setup guide. It now utilizes Bedrock Agents.

Released Java SDK v3.1.0

Released v3.1.0 of the Pinecone Java SDK. This version introduces support for specifying a base URL for control and data plane operations.

Pinecone Assistant: Context snippets and structured data files

You can now retrieve the context snippets that Pinecone Assistant uses to generate its responses. This data includes relevant chunks, relevancy scores, and references.

You can now upload JSON (.json) and Markdown (.md) files to an assistant.

Monthly spend alerts

You can now set an organization-wide monthly spend alert. When your organization’s spending reaches the specified limit, you will receive an email notification.

Released .NET SDK v2.0.0

Released v2.0.0 of the Pinecone .NET SDK. This version uses the latest stable API version, 2024-10, and adds support for embedding, reranking, and import. It also adds support for using the .NET SDK with proxies.

Released Python SDK v5.4.0 and v5.4.1

Released v5.4.0 and v5.4.1 of the Pinecone Python SDK. v5.4.0 adds a query_namespaces utility method to run a query in parallel across multiple namespaces in an index and then merge the result sets into a single ranked result set with the top_k most relevant results. v5.4.1 adds support for the pinecone-plugin-inference package required for some integrated inference operations.

Enabled CSV export of usage and costs

You can now download a CSV export of your organization’s usage and costs from the Pinecone console.

Added Support chat in the console

You can now chat with the Pinecone support bot and submit support requests directly from the Pinecone console.

Published Assistant quickstart guide

Added an Assistant quickstart.

Cequence released updated Scala SDK

Cequence released a new version of their community-supported Scala SDK for Pinecone. See their blog post for details.

Added index tagging for categorization

You can now add index tags to categorize and identify indexes.

Released major SDK updates: Node.js, Go, Java, and Python

Released v4.0.0 of the Pinecone Node.js SDK. This version uses the latest stable API version, 2024-10, and adds support for reranking and import.

Released v2.0.0 of the Pinecone Go SDK. This version uses the latest stable API version, 2024-10, and adds support for reranking and import.

Released v3.0.0 of the Pinecone Java SDK. This version uses the latest stable API version, 2024-10, and adds support for embedding, reranking, and import.

v5.3.0 and v5.3.1 of the Pinecone Python SDK use the latest stable API version, 2024-10. These versions were release previously.

Pinecone API version 2024-10 is now the latest stable version

2024-10 is now the latest stable version of the Database API and Inference API. For highlights, see SDKs below.

Pinecone Inference now available on the free Starter plan

The free Starter plan now supports reranking documents with Pinecone Inference.

Customer-managed encryption keys (CMEK) in early access

You can now use customer-managed encryption keys (CMEK) to secure indexes within a Pinecone project. This feature is in early access.

Serverless index monitoring generally available

Monitoring serverless indexes with Prometheus or Datadog is now in general availability.

Data import from Amazon S3 in public preview

You can now import data into an index from Amazon S3. This feature is in public preview.

Chat and update features added to Assistant

Added the chat_assistant endpoint to the Assistant API. It can be used to chat with your assistant, and get responses and citations back in a structured form.

You can now add instructions when creating or updating an assistant. Instructions are a short description or directive for the assistant to apply to all of its responses. For example, you can update the instructions to reflect the assistant’s role or purpose.

You can now update an existing assistant with new instructions or metadata.

Added the Matillion integration page.

Added guidance on using the Node.js SDK with proxies.

Released v5.3.1 of the Pinecone Python SDK. This version adds a missing python-dateutil dependency.

Released v1.1.1 of the Pinecone Go SDK. This version adds support for non-secure client connections.

Released v2.1.0 of the Pinecone Java SDK. This version adds support for non-secure client connections.

Released v5.3.0 of the Pinecone Python SDK. This version adds support for import operations. This feature is in public preview.

Added the metrics_alignment operation, which provides a way to evaluate the correctness and completeness of a response from a RAG system. This feature is in public preview.

When using Pinecone Assistant, you can now choose an LLM for the assistant to use and filter the assistant’s responses by metadata.

Added the Datavolo integration pages.

Released v5.2.0 of the Pinecone Python SDK. This version adds support for reranking documents with Pinecone Inference; it is no longer necessary to install the pinecone-plugin-inference package seperately. This feature is in public preview.

Prometheus monitoring for serverless indexes is now in public preview.

Released v3.0.3 of the Pinecone Node.js SDK. This version removes extra logging and makes general internal enhancements.

If you are upgrading from the Starter plan, you can now connect your Pinecone organization to the AWS Marketplace, GCP Marketplace, or Azure Marketplace for billing purposes.

Refreshed the navigation and overall visual interface of the Pinecone console.

Added Go examples for batch upserts, parallel upserts, and deleting all records for a parent document.

Added the Aryn integration page.

Released v3.0.2 of the Pinecone Node.js SDK. This version removes a native Node utility function that was causing issues for users running in Edge. There are no downstream affects of its removal; existing code should not be impacted.

Released v5.1.0 of the Pinecone Python SDK. With this version, the SDK can now be installed with pip install pinecone / pip install "pinecone[grpc]". This version also includes a has_index() helper function to check if an index exists.

Released v0.1.0 and v0.1.1 of the Pinecone Rust SDK. The Rust SDK is in “alpha” and is under active development. The SDK should be considered unstable and should not be used in production. Before a 1.0 release, there are no guarantees of backward compatibility between minor versions. See the Rust SDK README for full installation instructions and usage examples.

Released v1.0.0 of the Pinecone .NET SDK. For usage examples, see our guides or the GitHub README.

You can now back up and restore serverless indexes. This feature is in public preview.

Serverless indexes are now in general availability on GCP and Azure for Standard and Enterprise plans.

You can now deploy serverless indexes in the europe-west1 (Netherlands) region of GCP.

Released v1.1.0 of the Pinecone Go SDK. This version adds support for generating embeddings via Pinecone Inference.

Added the Nexla integration page.

Pinecone Assistant is now in public preview.

The Pinecone Inference API now supports reranking. This feature is in public preview.

Released v1.0.0 of the Pinecone Go SDK. This version depends on Pinecone API version 2024-07 and includes the ability to prevent accidental index deletion. With this version, the Go SDK is officially supported by Pinecone.

Added the Nuclia integration page

Added the Redpanda integration page.

Updated the Build a RAG chatbot guide to use Pinecone Inference for generating embeddings.

Added the ability to prevent accidental index deletion.

Released v5.0.0 of the Pinecone Python SDK. This version depends on Pinecone API version 2024-07 and includes the ability to prevent accidental index deletion. Additionally, the pinecone-plugin-inference package required to generate embeddings with Pinecone Inference is now included by default; it is no longer necessary to install the plugin seperately.

Released v3.0.0 of the Pinecone Node.js SDK. This version depends on Pinecone API version 2024-07 and includes the ability to prevent accidental index deletion. Additionally, this version supports generating embeddings via Pinecone Inference.

Released v2.0.0 of the Pinecone Java SDK. This version depends on Pinecone API version 2024-07 and includes the ability to prevent accidental index deletion. Additionally, this version includes the following breaking changes:

• createServerlessIndex() now requires a new argument: DeletionProtection.ENABLED or DeletionProtection.DISABLED.
• configureIndex() has been renamed configurePodsIndex().

For more details, see the Java SDK v2.0.0 migration guide.

Released version 2024-07 of the Database API and Inference API. This version includes the following highlights:

• The create_index and configure_index endpoints now support the deletion_protection parameter. Setting this parameter to "enabled" prevents an index from accidental deletion. For more details, see Prevent index deletion.

• The describe_index and list_index responses now include the deletion_protection field. This field indicates whether deletion protection is enabled for an index.

• The spec.serverless.cloud and spec.serverless.region parameters of create_index now support gcp / us-central and azure / eastus2 as part of the serverless public preview on GCP and Azure.

Serverless indexes are now in public preview on Azure for Standard and Enterprise plans.

Released version 1.1.0 of the official Spark connector for Pinecone. In this release, you can now set a source tag. Additionally, you can now upsert records with 40KB of metadata, increased from 5KB.

Serverless indexes are now in public preview on GCP for Standard and Enterprise plans.

Added an introduction to key concepts in Pinecone and how they relate to each other.

Added the Twelve Labs integration page.

Added a model gallery with details and guidance on popular embedding and reranking models, including models hosted on Pinecone’s infrastructure.

Released version 1.2.2 of the Pinecone Java SDK. This release simplifies the proxy configuration process. It also fixes an issue where the user agent string was not correctly setup for gRPC calls. Now, if the source tag is set by the user, it is appended to the custom user agent string.

You can now load a sample dataset into a new project.

Simplified the process for migrating paid pod indexes to serverless.

The Assistant API is now in beta release.

The Inference API is now in public preview.

Added a new legal semantic search sample app that demonstrates low-latency natural language search over a knowledge base of legal documents.

Added the Instill integration page.

Added the Langtrace integration page.

Updated Python code samples to use the gRPC version of the Python SDK, which is more performant than the Python SDK that interacts with Pinecone via HTTP requests.

Released version 4.1.1 of the Pinecone Python SDK. In this release, you can now use colons inside soure tags. Additionally, the gRPC version of the Python SDK now allows retries of up to MAX_MSG_SIZE.

The Enterprise quota for namespaces per serverless index has increased from 50,000 to 100,000.

Added the Fleak integration page.

Released version 1.2.1 of the Pinecone Java SDK. This version fixes the error Could Not Find NameResolverProvider using uber jar.

Added the Gathr integration page.

Released version 1.1.0 of the Pinecone Java SDK. This version adds the ability to list record IDs with a common prefix.

Released version 1.2.0 of the Pinecone Java SDK. This version adds the ability to list all record IDs in a namespace.

Added the following integration pages:

• Apify
• Context Data
• Estuary
• GitHub Copilot
• Jina
• FlowiseAI
• OctoAI
• Streamnative
• Traceloop
• Unstructured
• VoyageAI

You can now use the ConnectPopup function to bypass the Connect widget and open the “Connect to Pinecone” flow in a popup. This can be used in an app or website for a seamless Pinecone signup and login process.

Released version 1.0.0 of the official Spark connector for Pinecone. In this release, you can now upsert records into serverless indexes.

Pinecone now supports AWS PrivateLink. Create and use Private Endpoints to connect AWS PrivateLink to Pinecone while keeping your VPC private from the public internet.

Released version 4.0.0 of the Pinecone Python SDK. In this release, we are upgrading the protobuf dependency in our optional grpc extras from 3.20.3 to 4.25.3. Significant performance improvements have been made with this update. This is a breaking change for users of the optional GRPC addon (installed with pinecone[grpc]).

• The docs now have a new AI chatbot. Use the search bar at the top of our docs to find related content across all of our resources.
• We’ve updated the look and feel of our example notebooks and sample apps. A new sample app, Namespace Notes, a simple multi-tenant RAG app that uploads documents, has also been added.

The free Starter plan now includes 1 project, 5 serverless indexes in the us-east-1 region of AWS, and up to 2 GB of storage. Although the Starter plan has stricter limits than other plans, you can upgrade whenever you’re ready.

Pinecone now provides a Connect widget that can be embedded into an app, website, or Colab notebook for a seamless signup and login process.

Added the lifecycle policy of the Pinecone API, which describes the availability phases applicable to APIs, features, and SDK versions.

As announced in January 2024, control plane operations like create_index, describe_index, and list_indexes now use a single global URL, https://api.pinecone.io, regardless of the cloud environment where an index is hosted. This is now in general availability. As a result, the legacy version of the API, which required regional URLs for control plane operations, is deprecated as of April 15, 2024 and will be removed in a future, to be announced, release.

Added the Terraform integration page.

Released version 0.9.0 of the Canopy SDK. This version adds support for OctoAI LLM and embeddings, and Qdrant as a supported knowledge base. See the v0.9.0 release notes in GitHub for more details.

You can now deploy serverless indexes in the eu-west-1 region of AWS.

Released version 1.0.0 of the Pinecone Java SDK. With this version, the Java SDK is officially supported by Pinecone. For full details on the release, see the v1.0.0 release notes in GitHub. For usage examples, see our guides or the GitHub README. To migrate to v1.0.0 from version 0.8.x or below, see the Java v1.0.0 migration guide.

Added a Troubleshooting section, which includes content on best practices, troubleshooting, and how to address common errors.

Added an explanation of the Pinecone serverless architecture, including descriptions of the high-level components and explanations of the distinct paths for writes and reads.

Added considerations for querying serverless indexes with metadata filters.

Released version 3.2.2 of the Pinecone Python SDK. This version fixes a minor issue introduced in v3.2.0 that resulted in a DeprecationWarning being incorrectly shown to users who are not passing in the deprecated openapi_config property. This warning can safely be ignored by anyone who is not preparing to upgrade.

Released version 3.2.0 of the Pinecone Python SDK. This version adds four optional configuration properties that enable the use of Pinecone via proxy.

Released version 2.2.0 of the Pinecone Node.js SDK. This releases adds an optional sourceTag that you can set when constructing a Pinecone client to help Pinecone associate API activity to the specified source.

Released version 0.4.1 of the Pinecone Go SDK. This version adds an optional SourceTag that you can set when constructing a Pinecone client to help Pinecone associate API activity to the specified source.

Released version 2.2.0 of the Pinecone Node.js SDK.

Released version 0.4.1 of the Pinecone Go SDK.

Released version 3.2.1 of the Pinecone Python SDK. This version adds an optional source_tag that you can set when constructing a Pinecone client to help Pinecone associate API activity to the specified source. See the v3.2.1 release notes in GitHub for more details.

Released version 0.8.1 of the Canopy SDK. This version includes bug fixes, the removal of an unused field for Cohere chat calls, and added guidance on creating a knowledge base with a specified record encoder when using the core libary. See the v0.8.1 release notes in GitHub for more details.

The Pinecone console has a new look and feel, with a brighter, minimalist design; reorganized menu items for quicker, more intuitive navigation; and easy access to recently viewed indexes in the sidebar.

When viewing the list of indexes in a project, you can now search indexes by index name; sort indexes alphabetically, by how recently they were viewed or created, or by status; and filter indexes by index type (serverless, pod-based, or starter).

Released version 0.4.0 of the Pinecone Go SDK. This version is a comprehensive re-write and adds support for all current Pinecone API operations.

Fixed a bug that caused inaccurate index fullness reporting for some pod-based indexes on GCP.

You can now deploy serverless indexes in the us-east-1 region of AWS.

Released version 2.1.0 of the Pinecone Node.js SDK. This version adds support for listing the IDs of records in a serverless index. You can list all records or just those with a common ID prefix. Listing by common ID prefix is especially useful as part of managing RAG documents.

You can now configure single single-on to manage your teams’ access to Pinecone through any identity management solution with SAML 2.0 support, such as Okta. This feature is available on the Enterprise plan only.

Updated the Langchain integration guide to avoid a namespace collision issue.

The latest version of the Canopy SDK (v0.8.0) adds support for Pydantic v2. For applications depending on Pydantic v1, this is a breaking change; review the Pydantic v1 to v2 migration guide and make the necessary changes before upgrading. See the Canopy SDK release notes in GitHub for more details.

The latest version of Pinecone’s Python SDK (v3.1.0) adds support for listing the IDs of records in a serverless index. You can list all records or just those with a common ID prefix. Listing by common ID prefix is especially useful as part of managing RAG documents. See the Python SDK release notes in GitHub for more details.

Improved the docs for setting up billing through the AWS marketplace and GCP marketplace.

It is now possible to convert a pod-based starter index to a serverless index. For organizations on the Starter plan, this requires upgrading to Standard or Enterprise; however, upgrading comes with $100 in serverless credits, which will cover the cost of a converted index for some time.

Added a Llamaindex integration guide on building a RAG pipeline with LlamaIndex and Pinecone.

The latest version of the Canopy SDK (v0.6.0) adds support for the new API mentioned above as well as namespaces, LLMs that do not have function calling functionality for query generation, and more. See the release notes in GitHub for more details.

The latest versions of Pinecone’s Python SDK (v3.0.0) and Node.js SDK (v2.0.0) support the new API. To use the new API, existing users must upgrade to the new client versions and adapt some code. For guidance, see the Python SDK v3 migration guide and Node.js SDK v2 migration guide.

The Pinecone documentation is now versioned. The default “latest” version reflects the new Pinecone API. The “legacy” version reflects the previous API, which requires regional URLs for control plane operations and does not support serverless indexes.

The new Pinecone API gives you the same great vector database but with a drastically improved developer experience. The most significant improvements include:

• Serverless indexes: With serverless indexes, you don’t configure or manage compute and storage resources. You just load your data and your indexes scale automatically based on usage. Likewise, you don’t pay for dedicated resources that may sometimes lay idle. Instead, the pricing model for serverless indexes is consumption-based: You pay only for the amount of data stored and operations performed, with no minimums.

• Multi-region projects: Instead of choosing a cloud region for an entire project, you now choose a region for each index in a project. This makes it possible to consolidate related indexes in the same project, even when they are hosted in different regions.

• Global URL for control plane operations: Control plane operations like create_index, describe_index, and list_indexes now use a single global URL, https://api.pinecone.io, regardless of the cloud environment where an index is hosted. This simplifies the experience compared to the legacy API, where each environment has a unique URL.