Snowpipe Streaming SDK release notes for 2026¶

Version 1.5.0 (May 20, 2026)¶

New features and updates¶

• Added the application client property and SS_APPLICATION environment variable for partner CSID attribution. Partners and integrators can now identify their application to Snowflake for usage tracking and support. For more information, see Configurations for Snowpipe Streaming with high-performance architecture.
• The SDK now uses the jemalloc memory allocator on Linux and macOS to reduce memory fragmentation under sustained streaming workloads. The change is automatic; no configuration is required. Windows builds continue to use the system allocator. For more information, see Best practices for Snowpipe Streaming with high-performance architecture.
• Enabled FIPS 140-3 validated cryptography for TLS and data encryption. The SDK now uses AWS-LC as its cryptographic provider and enforces HTTPS-only connections to Snowflake. For more information, see Configurations for Snowpipe Streaming with high-performance architecture.
• Added support for running the SDK inside Snowpark Container Services (SPCS). The SDK reads the SPCS workload identity token automatically and authenticates without long-lived credentials in the container. For more information, see Run the Snowpipe Streaming SDK in Snowpark Container Services.
• Released the Snowpipe Streaming SDK for Node.js v1.5.0 on npm with the same updates as the Java and Python SDKs.

For links to the official SDK packages for each supported language, see How to connect.

Version 1.4.0 (May 08, 2026)¶

New features and updates¶

• The Snowpipe Streaming SDK for Node.js is now generally available. The SDK brings the Snowpipe Streaming high-performance architecture to JavaScript and TypeScript applications, and shares the same Rust-based client core as the Java and Python SDKs. Install from npm, and see the Node.js API reference. The Node.js SDK requires Node.js 20 or later and runs on ARM64 Mac, Windows, ARM64-Linux, and x86_64-Linux (glibc 2.26 or later).
• Added support for OAuth authentication. You can now set authorization_type to OAUTH and connect to Snowflake using OAuth (Snowflake OAuth or External OAuth, including the refresh-token and client-credentials flows). For configuration details, see Configurations for Snowpipe Streaming with high-performance architecture.
• Added support for programmatic access token (PAT) authentication. You can now set authorization_type to PAT and provide a personal_access_token to connect to Snowflake using a long-lived programmatic access token.
• Binary column encoding now follows the effective BINARY_INPUT_FORMAT parameter for custom pipes. The SDK fetches the effective format through the pipe-info API at client creation and encodes byte arrays accordingly. Default pipes always use BASE64 for binary columns, independent of account-level or session-level parameters. If the effective BINARY_INPUT_FORMAT changes during a session, the SDK invalidates the client and you must close and reopen it. For more information, see The PIPE object.
• The appendRows method now accepts a startOffsetToken in addition to the existing end offset token, letting you record the offset range covered by each batch. For single-row appends, the SDK uses the same token for both bounds. The start offset is recorded as offset_token_lower_bound in error-table metadata, making it easier to locate the source range that produced an erroneous row. For more information, see Error logging in Snowpipe Streaming with high-performance architecture.

Version 1.3.0 (March 12, 2026)¶

New features and updates¶

• Maintenance and stability release. Updates to release tooling and minor internal improvements. See the Maven Central release page for the artifact.

Version 1.2.0 (February 16, 2026)¶

New features and updates¶

• Added support for encrypted key-pair authentication. You can now connect to your instances by using both encrypted and unencrypted private keys, providing greater flexibility for your security workflows.

Version 1.1.2 (January 20, 2026)¶

Behavior changes¶

• Fixed a race condition in the channel status cache to improve multi-threaded stability.
• Reduced log flooding by removing redundant messages for cleaner monitoring.

New features and updates¶

• Added support for account locators with region suffixes.

Bug fixes¶

• Removed unused configuration parameters and addressed minor internal logic issues to improve reliability.