augusto kiniama rosa

Posted on Oct 21 • Originally published at blog.archetypeconsulting.com on Jun 3

The Unofficial Snowflake Monthly Release Notes: May 2025 (includes Summit 2025 Announcements)

#data #snowflake #newfeatures #ai

Monthly Snowflake Unofficial Release Notes #New features #Previews #Clients #Behavior Changes

Welcome to the Unofficial Release Notes for Snowflake for May 2025! It includes many of the releases announced during the Snowflake Summit 2025 but released into the platform during May. You’ll find all the latest features, drivers, and more in one convenient place.

As an unofficial source, I am excited to share my insights and thoughts. Let’s dive in! You can also find all of Snowflake’s releases here.

This month, we provide coverage up to release 9.13 (General Availability — GA). I hope to extend this eventually to private preview notices as well.

I would appreciate your suggestions on continuing to combine these monthly release notes. Feel free to comment below or chat with me on LinkedIn.

Behavior change bundle 2025_01 is generally enabled for all customers, 2025_02 is enabled by default but can be opted out until next BCR deployment, and 2025_03 is disabled by default but may be opted in.

What’s New in Snowflake

New Features

Release channels for Snowflake Native Apps (GA) enable providers to publish apps at various development stages. For instance, a provider can use release channels to: Test an app, Publish an app for consumer preview or UAT (user acceptance testing), and Publish the app to production.
Snowpark Python version updates (support for Python version 3.12)
Data types: Structured types support for standard Snowflake tables (Preview), they are ARRAY, OBJECT, and MAP
Snowflake Openflow (Preview) connects any data source to any destination with processors for structured/unstructured text, images, audio, video, and sensor data. Built on Apache NiFi, Openflow provides a fully managed service in your cloud for complete control
Cost anomalies (Preview), automatically identify cost irregularities based on previous consumption levels. This simplifies spotting cost spikes or drops, allowing you to find opportunities for optimizing spending. This feature enables you to detect cost anomalies at both the account and organization levels
Generation 2 standard warehouses (GA) is an updated version (the “next generation”) of Snowflake's current standard virtual warehouse, aiming to enhance performance for analytics and data engineering workloads. Built on faster hardware and smart software optimizations, like improved delete, update, merge operations, and table scans, Gen2 enables quicker query completions and simultaneous task handling with some pretty good early results at 2.1x speed improvements but they do cost a little more

Snowsight Updates

New Snowsight navigation (Preview): updated navigation experience in Snowsight is available, Work with data: Ingest, transform, analyze, and monitor data using integrated tools for development and automation, Discover & collaborate: Search, share, and manage data across teams and partners with cataloging and Marketplace features, and Manage: Control access, optimize compute, and maintain governance across your Snowflake environment
Universal Search: support for pipes, tasks, and streams (GA), Universal Search capabilities have been expanded to include pipes, tasks, and streams

AI Updates (Cortex, ML, DocumentAI)

LLaMA 4 models support for COMPLETE multimodal (Preview), supports multimodal models from Meta and Anthropic, specifically the following: Claude 3.7 Sonnet, Claude 4 Sonnet, Claude 4 Opus, LLaMA 4 Maverick, and LLaMA 4 Scout
JSON schema references for COMPLETE Structured Output: Provide JSON schema references to ensure structured outputs from the Cortex COMPLETE function
Snowflake supports schema references in Cortex COMPLETE Structured Outputs, simplifying the creation and maintenance of complex schemas. The new $ref mechanism lets developers define common components once and reference them throughout their schema. This enhancement also ensures compatibility with third-party libraries like Pydantic, allowing the use of existing Pydantic schemas with COMPLETE Structured Outputs
Table extraction in Document AI (Preview), extract tables from documents using the new Snowflake Arctic-TILT foundation model geared towards tabular and relational extraction
Snowflake Cortex Provisioned Throughput (GA): Ensure predictable performance for Cortex AI workloads with provisioned throughput. Use it to reserve throughput for specific periods with provisioned throughput units (PTUs), allocate capacity for supported models, and scale throughput based on workload needs with minimum and incremental configurations
Snowflake Copilot model RBAC enables role-based access control (RBAC) for managing which large language models (LLMs) can be used by Snowflake Copilot and other Cortex features according to user role
Snowflake ML Data Connector for Container Runtime (GA), allows efficient data ingestion from Snowflake into containerized Python environments, leveraging distributed processing to speed up loading
Document AI updates, a new version of the foundational Arctic-TILT model in Document AI includes improvements in checkbox identification

Snowflake Applications (Container Services, Notebooks and Applications)

Snowpark Container Services preview available in Google Cloud (Preview): You can now leverage Snowpark Container Services in Google Cloud environments
Notebooks Container Runtime on Azure and Azure Private Link: enhancements for running Snowflake Notebooks with container runtimes on Azure, including Private Link support
Data Connector for Container Runtime (GA): Snowflake ML Data Connector for Container Runtime is now generally available
Using caller’s rights to connect to Snowflake from Snowpark Container Services (GA): Connect to Snowflake from within a container using the caller’s rights for enhanced security and context
Notebooks st.secrets support for Warehouse and Container Runtimes: Manage secrets in Snowflake Notebooks more effectively across different runtimes
Snowpark Container Services preview available in all Google Cloud (Preview)
Streamlit, Support for Streamlit 1.44.0 (GA)

Data Lake Updates

Apache Iceberg™ tables: Row-level deletions are now available for externally managed tables (GA), this feature supports row-level deletes with positional delete files, enabling external engines to carry out update, delete, and merge operations on Iceberg tables managed in Snowflake
Query Acceleration Service: Support for Apache Iceberg™ tables (QAS accelerates scan performance and inserts on Iceberg tables)
Search optimization: Support for Apache Iceberg tables (improve the performance of queries on Iceberg tables)
Support for Iceberg tables in the People’s Republic of China (GA): Apache Iceberg™ tables are now generally available for use in the People’s Republic of China
Cross-cloud/cross-region support for Snowflake-managed Apache Iceberg™ tables (GA): supports cross-cloud and cross-region reads and writes for Iceberg tables that use Snowflake as the catalog. This includes the ability to convert tables using an external catalog to use Snowflake’s catalog
INFER_SCHEMA function: Support for Apache Iceberg™ data types (GA): The INFER_SCHEMA function can now automatically retrieve column definitions for Apache Iceberg™ data types from staged files, facilitating the creation of Iceberg tables using templates

Data Pipelines/Data Loading/Unloading Updates

Triggered tasks: Utilize Triggered Tasks to execute stored procedures in response to changes in streams hosted on directory tables or data shares
Support for internal stage cloning (GA), support for internal stage cloning when you clone a database or schema
Vectorized scanner now available without ON_ERROR restrictions, allowing you to leverage its performance benefits regardless of your chosen ON_ERROR setting, including CONTINUE, SKIP_FILE_num, and ‘SKIP_FILE_num%’. This offers greater flexibility in configuring data loading processes while optimizing scanning

Security, Privacy & Governance Updates

Organization users (Preview): Multi-account organizations can now create an organization user for a person who needs to access multiple accounts
Sensitive data classification: New classifiers for India (GA): Snowflake’s sensitive data classification now supports new classifiers for India, including NATIONAL_IDENTIFIER (covering Permanent Account Number (PAN), Aadhaar, and Voter ID), DRIVERS_LICENSE, and TAX_IDENTIFIER (Goods and Service Tax Identification Number (GSTIN))
Outbound private connectivity for AWS Government regions (GA): AWS PrivateLink can now be used for outbound network traffic originating from Snowflake accounts in AWS Government regions
Trust Center: In-app notifications (Preview), receive Trust Center notifications in Snowsight about accounts that haven’t enrolled in multi-factor authentication (MFA)
Object tags available in Standard Edition, all accounts can now create and set object tags regardless of the account’s edition
New multi-factor authentication (MFA) methods allow users to authenticate using: a passkey stored in various ways (e.g., hardware security key or laptop fingerprint sensor), or an authenticator app generating a time-based one-time passcode (TOTP). Administrators can restrict available MFA methods during setup
Contacts for objects enable users to associate contacts with databases and tables for assistance. Each contact is a schema-level object detailing communication methods, such as email or URL access. An object can have multiple contacts with distinct purposes; for instance, a table might have one contact for access approval and another for general support
Automatically propagate user-defined tags (GA) by configuring an object tag to flow from source to target objects. This streamlines tag management and ensures consistent application of data protection policies on targets

SQL, Extensibility & Performance Updates

Enhanced error messages for Data Manipulation Language (DML) commands will now incorporate the column name in some cases
Search optimization: Support for Apache Iceberg tables (improve the performance of queries on Iceberg tables)
The pipe operator (->>) chains SQL statements, allowing the result of one to serve as input for another. It simplifies execution of dependent statements and enhances readability and flexibility in complex SQL operations
New SQL functions (Preview): DATASKETCHES_HLL, returns an approximation of the distinct cardinality of the input (that is, DATASKETCHES_HLL(col1) returns an approximation of COUNT(DISTINCT col1)); DATASKETCHES_HLL_ACCUMULATE, returns the sketch at the end of aggregation; DATASKETCHES_HLL_COMBINE, combines (merges) input sketches into a single output sketch; DATASKETCHES_HLL_ESTIMATE, returns the cardinality estimate for the given sketch
Built-in code profiler for Python stored procedures (GA), built-in code profiling for stored procedure handler code written in Python
Dynamic tables: Support for filtering by current time and date for incremental refresh — General availability We are pleased to announce support for CURRENT_TIMESTAMP, CURRENT_DATE, and CURRENT_TIME functions as filters for dynamic tables in incremental refresh mode. You can now use these functions in predicates like WHERE/HAVING/QUALIFY clauses
Dynamic tables now include support for IS_ROLE_IN_SESSION in access policies (GA), as well as base tables that implement row access or masking policies utilizing the IS_ROLE_IN_SESSION function for both incremental and full refresh modes
Improved performance of dynamic table refreshes that contain top-level QUALIFY clauses with RANK or ROW_NUMBER ranking window functions, specifically when the rank value is 1, using QUALIFY RANK() = 1 or ROW_NUMBER = 1 now refresh more quickly, improving performance for common deduplication and top-N use cases.

Marketplace, Listings & Data Sharing

Request Approval Workflow (GA), allows consumers to request access to Internal Marketplace organizational listings from Snowsight, and allows providers to view and manage access requests to organization listings in Snowsight
Data sharing and collaboration for accounts in the Kingdom of Saudi Arabia allow customers using Snowflake’s Data Cloud in the Middle East to discover and securely share data and execute various analytic workloads, utilizing Snowflake’s implementation on Azure UAE North in Dubai
Organizational listings: With this, administrators can define which accounts and roles can access listings. Listing owners can select the entire organization, specific accounts, or roles to control access. Creators can specify access in the Internal Marketplace using the same options

Data Clean Rooms Updates

Free-form SQL queries are now available in the API, exposing your linked tables and views in a clean room to be available for free-form queries by clean room collaborators in any Snowflake coding environment
Sign in with Snowflake Authentication: You can sign in to the clean rooms UI using your Snowflake credentials. The process integrates smoothly with your Snowflake SSO/SAML
Activation methods have been introduced for run roles. Users with run roles can now activate their data. Check the complete list of procedures accessible to run role users
Differential privacy is now managed, reducing costs for clean room owners. This feature will automatically enable for new clean room accounts
Providers can choose the warehouse size for decrypting and storing activation results by calling provider.update_activation_warehouse

Open-Source Updates

terraform-snowflake-provider 2.1.0 (upgrade to the GoSnowflake driver, Several adjustments and fixes were made to the internal model builders, including ordering fixes, adjustments for tag associations, generation of dynamic blocks, and handling of forbidden attribute names. These changes contribute to the stability and maintainability of the provider’s codebase, various documentation updates were made, including changes to date formatting, removal of notes regarding future Streamlits, documentation for deprecated grant resource mapping, and generation of stable and preview resources in the documentation, removal of a tautological err == nil check in a network policy attachment test, Correction in setting values into the correct field for field transformers, Fixes for tests that were expected to fail once the 02_2025 Snowflake bundle is enabled, Manual import of a GPG key, Improved masking of random values, Adjustments to environment variable usage for providing configuration files in CI/CD pipelines, adoption of the new TOML format in pipelines, u8pdates to automatic object tracking versioning)
Modin 0.33.1 ( Bug fixes: add copy parameter to array methods)
Modin 0.33.0 ( New features: Replace NativeDataFrameMode with a complete “native” execution, Add metrics interface so third-parties can collect metrics from the modin frontend, Allow QueryCompilerCaster to apply cost-optimization on automatic casting, Add Backend config variable as an alias for execution, Add methods to get and set backend, Add progress bar for engine switch, Add an option register dataframe and series accessors with a particular backend, Register general functions with a particular backend, Choose the correct init method from extensions and apply casting to init, Move the query compiler calculator so it can be used in more places, Implement max_cost interface, Add “from_qc” API to QueryCompiler and BackendCostCalculator to handle asymmetric information scenarios, Allow I/O function accessors, Support post-operation automatic backend switch, Support pre-operation automatic backend switch, Add AutoSwitchBackend configuration variable, Support pre-operation switch for init by passing arguments to cost functions, Support pinning objects to a backend, Improve formal definition of the automatic switching algorithm, Ability to configure NativeQueryCompiler AutoSwitch Settings, Support post-operation backend switch for groupby, Let plugins register groupby accessors, Emit metrics on auto-switch and casting behavior, Add operation and size information to backend switch progress, Dispatch array_ufunc to query compilers, bug fixes: Corrected Series.fillna with method="bfill" and method="ffill" when limit is specified to ensure correct limited filling behavior,Fixed an issue in DataFrame.sort_values with ignore_index=True on an empty DataFrame to prevent errors,Resolved a bug in DataFrame.mode with dropna=False for object dtype columns to correctly handle NaN values,Addressed incorrect warning message for read_csv_glob when using usecols parameter,Fixed Series.value_counts to correctly handle normalize=True when the Series contains only NaN values,Ensured Series.value_counts output is correctly named when normalize=True,Fixed DatetimeProperties.isocalendar to correctly handle timezones by converting to UTC first,Corrected the dtypes of result in pivot_table when margins_name is not default for consistent type output,Resolved an issue in Series.plot when use_index=False and the index is a MultiIndex to correctly plot data without the index,Fixed Series.fillna to handle limit parameter correctly when the Series is empty,Corrected an issue with read_csv type inference when encoding="utf-16" leading to incorrect dtypes,Fixed DataFrame.loc with a boolean Series indexer that has a different index than the DataFrame being indexed,Addressed an issue with DataFrame.merge when merging on mixed-type columns where one is object and the other is numeric,Fixed a bug in DataFrame.iterrows when used with certain data types that caused incorrect iteration,Resolved an issue in DataFrame.apply with axis=1 where args were not correctly passed to the applied function in some scenarios,Fixed a bug in DataFrame.quantile when dealing with empty DataFrames or all-NaN columns to align with pandas behavior, Corrected Series.resample().apply() when the custom aggregation function returned a list, Updated developer documentation regarding releasing Modin,Various internal code improvements and refactoring for stability and performance)
Snowflake VS Code Extension 1.14.0 (General Availability release of SnowConvert Migration Assistant, Resolve SnowConvert migration issues (EWIs, FDMs, PRFs) with AI-powered assistance using Snowflake Cortex AI, Multiple interaction methods: Access assistance through the Issues panel or CodeLens integration, Enhanced CodeLens integration with in-editor links for instant AI assistance on specific migration issues, Improved SnowConvert Issues panel to browse and navigate conversion issues in your workspace, Configurable AI model preferences with automatic fallback support for Claude 3.7 Sonnet, Claude 3.5 Sonnet, Claude 4 Sonnet, Llama 3.1 70B, and Mistral Large 2, Expanded chat interaction scope to answer any SQL-related question beyond migration-specific issues, Enable this feature in your settings via the snowflake.snowConvertMigrationAssistant.enabled property)
Snowflake VS Code Extension 1.13.3 (Snowpark Checkpoints: Added information properties to checkpoints.json file, Added interactive checkpoints-execution buttons, Improved “Load All Checkpoints” initial message)
Snowflake VS Code Extension 1.13.2 (Snowpark Checkpoints: Fixed checkpoints crash when a checkpoint could not be added)
Streamlit 1.45.1 (Bug Fix: Enhanced ETag Compatibility for st.video that improves the performance and reliability of video handling)

Client, Drivers, Libraries and Connectors Updates

New features:

.NET Driver 4.5.0 (Added OAuth 2.0 Authorization Code flow authentication: added the oauth_authorization_code authenticator, added the oauthScope, oauthClientId, oauthClientSecret, oauthAuthorizationUrl, oauthTokenRequestUrl, and oauthRedirectUri connection parameters to configure the authentication, added the ability to provide oauthClientSecret by setting the SnowflakeDbConnection.OAuthClientSecret property instead of providing it in a connection string, added a cache for OAuth 2.0 tokens, added OAuth 2.0 Client Credential flow authentication: added the oauth_client_credentials authenticator, added oauthScope, oauthClientId, oauthClientSecret, and oauthTokenRequestUrl connection parameters to configure the authentication, added the ability to provide oauthClientSecret by setting the SnowflakeDbConnection.OAuthClientSecret property instead of providing it in a connection string, added Programmatic Access Token authentication: added the programmatic_access_token authenticator, added the ability to specify the token parameter either in a connection string or by setting the SnowflakeDbConnection.Token property, added validations for the scheme, port, and host connection properties, added the ability to provide tokens by setting the SnowflakeDbConnection.Token property instead of providing them in a connection string)
Go Snowflake Driver 1.14.1 (added support for propagating OpenTelemetry contexts to GS, added support for default client credentials in the OAuth authorization code flow, moved OCSP initialization to the first HTTPS call.)
Node.js 2.1.0 (added support for OAuth 2.0 Authorization Code Flow and OAuth 2.0 Client Credentials Flow, for OAuth 2.0 Authorization Code Flow: added the oauthClientId, oauthClientSecret, oauthAuthorizationUrl, oauthTokenRequestUrl, and oauthScope parameters, added the OAUTH_AUTHORIZATION_CODE parameter for the parameter authenticator, For OAuth 2.0 Client Credentials Flow: added the oauthClientId, oauthClientSecret, oauthTokenRequestUrl, and oauthScope parameters, added the OAUTH_CLIENT_CREDENTIALS parameter for the parameter authenticator, added support for virtual-style domains)
JDBC Driver 3.24.1 (added the HttpHeadersCustomizer interface to provide a flexible way to inject custom HTTP headers into various requests initiated by the Snowflake JDBC driver, added the LOCAL_APPLICATION default for the clientId and clientSecret OAUTH parameters)
PHP PDO Driver for Snowflake 3.2.0 (added support multi-factor authentication (MFA))
Snowflake CLI 3.9.0 (added the — encryption option to the snow stage create command to define the type of encryption to use for all files on the stage)
Snowflake CLI 3.8.0 (added support for OAuth tokens, added the following enhancements to the snow sql command: added an interactive mode, added support for asynchronous SQL queries, added support for the !queries, !result, and !abort SQL query commands, added the — single-transaction command line option to execute multiple SQL queries as an all-or-nothing batch, ensuring that all commands complete successfully before any of the changes are committed, added the artifact_repository and artifact_repository_packages field to the Snowpark Entity Model to support using non-anaconda packages)
Snowflake Python API 1.5.0 (Deprecated the ServiceResource.get_service_status method in favor of the ServiceResource.get_containers method, added the extract option to the procedure.call method. Enabling this option causes the method to extract results from the returned payload, added support for mapping the VARIANT type in a stored procedure call)
Snowpark Library for Python 1.32.0 (Added support for Python 3.12 allowing developers to utilize the latest Python version, Introduced Session.custom_package_usage_config property enabling users to view custom package usage configuration for the session, Enhanced snowflake-snowpark-python[pandas] installation by automatically including modin and s3fs as dependencies simplifying setup for users of the pandas API, added support for pyarrow >= 14.0.0 and < 17.0.0 when using the pandas API,Added support for pandas >= 2.0.0 and < 3.0.0 when using the pandas API ensuring compatibility with newer versions of these core libraries for pandas API users, Snowpark pandas API Updates: added support for dict values in Series.str.get, Series.str.slice, and Series.str.__getitem__ (Series.str[...]), added support for DataFrame.to_html, added support for DataFrame.to_string and Series.to_string, added support for reading files from S3 buckets using pd.read_csv, make iceberg_config a required parameter for DataFrame.to_iceberg and Series.to_iceberg )
Snowpark ML 1.8.5 (ML Jobs new features: job decorator now has min_instances argument that makes a job wait for the specified number of workers to be ready before starting)
Snowpark ML 1.8.4 (New Model Registry features: automatically enable explainability for models that can be deployed to a warehouse, new Explainability features: new visualization functions in snowflake.ml.monitoring plot explanations in notebooks, support for categorical transforms in scikit-learn pipelines, new Modeling features: support categorical types for xgboost.DMatrix inputs in XGBoost models)
SnowSQL 1.4.1 (upgraded snowflake-connector-python to 3.15.0)
SnowSQL 1.4.0 (fixed an issue with the snowsql --version command failing when automatic upgrades are disabled (noup=False))

Bug fixes:

Go Snowflake Driver 1.14.1 (aligned scan types and actually returned types for NUMBERs, fixed an issue with nil dereferencing when an internal timeout happened (for instance for cloud provider call) when the original context was still valid, fixed an issue with nil dereferencing during time out or canceling context race, fixed encryption bugs where errors were never returned, fixed downcast smkId to int, which caused decryption problems for very large stages, fixed support for virtual style domains on GCP, fixed the validation of the owner of the secure storage lock directory)
JDBC Driver 3.24.1 (fixed handling of timestamps before 04.10.1582 (Gregorian reform) when inserting with BindUploader, fixed NPE handling of writing to the cache file when the file is not accessible, fixed an issue with certificate chain traversal. The verification process now properly terminates upon reaching the root certificate, fixed the Workflow Identity Federation request signature for AWS)
PHP PDO Driver for Snowflake 3.2.0 (fixed a memory leak that occurred when fetching results, fixed an OCSP configuration issue)
Snowflake CLI 3.9.0 (fixed errors that occurred for use commands if the current database is not set)
Snowflake CLI 3.8.3 (added the --private-link option to the snow spcs image-registry url command for retrieving private link URLs)
Snowflake CLI 3.8.2 (changed the enable_release_channels property default from False to None)
Snowflake CLI 3.8.1 (the upgrade message is now sent to stderr, fixed a snowflake.core import issue on newer Python versions)
Snowflake CLI 3.8.0 (fixed an issue with deploying Snowpark project using the != operator in requirements.txt, fixed an issue with escaping identifiers for use commands, moved the enable_release_channels parameter from the global level to the project level, fixed the snow spcs service metrics command to accept fully qualified service names)
Snowflake CLI 3.7.2 (fixed an issue with errors appearing in help messages)
Snowflake Python API 1.5.1 (fixed a bug in ProcedureResource that caused the call method to return wrong results when using the extract option with the ReturnTable type,CortexInferenceService.complete can now be called from Python worksheets and notebooks)
Snowflake Python API 1.5.0 (fixed the type mapping for the GEOMETRY, GEOGRAPHY, OBJECT return types in stored procedures, the __repr__ implementation for stored procedures and functions now shows a list of arguments in addition to the name)
Snowpark Library for Python 1.32.0 (fixed an issue where StoredProcedureRegistration.register_from_file did not correctly handle relative paths when skip_upload_on_content_match was True improving reliability of stored procedure registration, corrected a problem where DataFrame.cache_result and DataFrame.uncache_result did not function properly when GEOGRAPHY (GEOGRAPHY) was enabled ensuring caching works as expected with geospatial data, addressed a bug where Session.sql().collect failed when statement_params was provided and GEOGRAPHY was enabled allowing SQL execution with parameters to work correctly with geospatial data, resolved an issue where Session.call did not handle None values correctly when GEOGRAPHY was enabled ensuring proper handling of nulls in stored procedure calls involving geospatial data)
Snowpark ML 1.8.5 (Behavior changes: ML Jobs behavior changes: Argument num_instances has been renamed to target_instances in job submission APIs and is now required, Model Registry bug fixes: Fixed a bug in listing and deleting container services, Fixed a bug with logging scikit-learn pipelines where the explain function was not created, Logging a container-only model no longer checks to make sure the required version of snowflake-ml-python is available in the Snowflake conda channel, Explainability bug fixes: Minimum streamlit version has been decreased to 1.30 to improve compatibility, Modeling bug fixes:xgboost is now a required dependency again (it was optional in v1.8.4))
Snowpark ML 1.8.4 (Behavior changes: ML Jobs behavior changes: the id property is now the job’s fully-qualified name. A new property, name, has been introduced to represent the ML Job name. the list_jobs method now returns the ML Job name instead of the job ID, model Registry behavior changes: In log_model, enabling explainability when the model is deployed only to Snowpark Container Services is now an error instead of a warning and will prevent the log operation from completing, bug fixes: Model Registry bug fixes: fixed a bug in which logging PyTorch and TensorFlow models that caused UnboundLocalError: local variable ‘multiple_inputs’ referenced before assignment)
SnowSQL 1.4.0 (added support for OAuth 2.0 Authorization Code Flow and OAuth 2.0 Client Credentials Flow, upgraded openssl to version 3.5.0, cryptography <= 44.0.3, updated how Windows binaries sign internally upgradable components)
Snowflake Connector for Google Analytics Raw Data 1.7.2 ()
Snowflake Connector for ServiceNow® 5.22.3 (fixed an issue with page size persistence when reloading a table)

Conclusion

The most recent developments from Snowflake show that it is still dedicated to pushing the limits of what is possible in the Data Cloud. Snowflake is clearly putting a lot of money into making advanced analytics more powerful and easier to use. For example, they are making Generation 2 standard warehouses available to everyone, which promise big performance boosts. They are also adding new LLaMA model support to Cortex, enhancing Document AI, and giving users more control with Copilot RBAC.

Key improvements in Data Lake technologies, especially the strong support and new GA features for Apache Iceberg tables, show a focus on open formats and flexibility. These include row-level deletions, the Query Acceleration Service, and support for multiple clouds and regions. With Snowpark Python 3.12 support, big changes to Snowpark Container Services that make them available on Google Cloud, and a more robust VS Code plugin with AI-assisted migration, developers have a better experience. Also, the preview of Snowflake Openflow suggests that data integration across different sources and destinations will be easier and more powerful in the future.

Improvements to Snowsight navigation and Universal Search, together with new features like cost anomaly detection and release channels for Native Apps, are meant to make the software easier to use, better at governance, and more efficient in its operations. There are also big improvements in security and governance, such as organization users, new sensitive data classifiers for India, more MFA choices, and more places where you can find object tags.

It’s amazing how many different things these updates cover, from core engine enhancements to AI/ML, developer tools, data lake capabilities, security, and open-source integrations like the updates to the Terraform provider, Modin, and numerous client drivers. All of these enhancements give businesses the power to build more, quicker, and with more confidence and control over their data. As Snowflake keeps coming up with new ideas at this fast rate, customers can look forward to a platform that is always changing to meet their needs and is ready for the challenges and opportunities of the future. It’s an exciting time to be working on Snowflake, and these new features provide you with a lot of new things to try out and use.

Please continue to share your feedback on how these unofficial notes can be improved. For the most detailed and authoritative information, always refer to the official Snowflake documentation.

Enjoy exploring these new capabilities!

I am Augusto Rosa, a Snowflake Data Superhero and Snowflake SME. I am also the Head of Data, Cloud, & Security Architecture at Archetype Consulting. You can follow me on LinkedIn.

Subscribe to my Medium blog https://blog.augustorosa.com for the most interesting Data Engineering and Snowflake news.

DEV Community