All release notes

• Apache Iceberg™ tables: Support for the UNKNOWN data type (Preview)

CoCo Desktop v1.20.0 focuses on a smoother, more reliable chat:

• Refreshed update card: The "ready to install" screen now matches the onboarding card design and includes a "What's new" link to the release notes.
• Better prompt caching and token efficiency: Improved prompt caching and tuning make sessions quicker and more cost-efficient on tokens.
• Draft messages are preserved: Unsent messages stick around across model swaps, editor hides, and session switches.

...

Read more

CoCo Desktop is now generally available. CoCo Desktop is a native Mac and Windows desktop application that brings CoCo's full agentic capabilities into an IDE-like surface, with a file editor, integrated terminal, agentic AI loop, agentic browser, agentic notebook, and deep Snowflake awareness.

...

Read more

Tag-based policy support is now available in public preview for aggregation policies, row access policies, projection policies, and join policies. This feature allows you to assign data protection policies to tags, automatically protecting any objects that have the tag assigned to them.

...

Read more

OpenID Connect (OIDC) security integrations for federated single sign-on (SSO) are available in Public Preview. Multiple OIDC integrations, or OIDC alongside an existing SAML2 integration, require identifier-first login. Without identifier-first login, only one SSO integration (SAML2 or OIDC) can be ENABLED=TRUE at a time.

...

Read more

Snowflake now maintains built-in network rules for five additional Microsoft partner applications:

• Microsoft Entra ID
• Microsoft Power Automate
• Microsoft Power Query Online

...

Read more

You can now use DML statements (INSERT, UPDATE, MERGE, DELETE) to modify rows directly in the frozen region of a dynamic table. This enables use cases such as GDPR-required row deletions or correcting specific records without triggering a full refresh of the active region.

...

Read more

CoCo Desktop added the following capabilities across releases v1.16.0 through v1.19.1:

• Personalization and Memory: A dedicated Personalization page for custom instructions and built-in tool toggles, plus persistent memory that carries context across conversations. For more information, see personalization and memory.
• Built-in web search and web fetch: The agent can search the web and fetch content from URLs when you enable those tools. For more information, see personalization.
• Semantic code search (tgrep): Find the right files in your codebase from a plain-English description, powered by Snowflake Cortex embeddings. For more information, see semantic code search.

...

Read more

DCM Projects now supports the following capabilities in preview:

• PLAN DELTA: A faster variant of PLAN that only evaluates definitions that changed since the last deployment, and any downstream definitions that depend on them. Use it during active development for faster feedback on incremental changes, but always run a full PLAN to review all changes before deploying.
• Python and Java functions and procedures: DEFINE FUNCTION and DEFINE PROCEDURE now support Python and Java handlers, in addition to SQL.
• ATTACH TAG: Declaratively assign Snowflake object tags, including to table and view columns, to any DCM Projects-managed entity. DCM Projects reconciles the declared tag assignments on every deployment.

...

Read more

The following new features and enhancements are now available in Snowflake Data Clean Rooms:

• Improved error handling in COLLABORATION.RUN: Previously, errors that occurred when using a hardcoded source table in template runs were incorrectly classified as replication lag issues. These errors are now correctly identified and reported.
• Improved validation in register_code_spec: When registering a Python artifact or ML job, the stage must use SNOWFLAKE_SSE encryption. The system now validates this at registration time and returns a clear error message, avoiding failures at runtime.
• General performance improvements and bug fixes.

VOLATILE scalar user-defined functions (UDFs) are now supported with dynamic table incremental refresh in INCREMENTAL, AUTO, and ADAPTIVE refresh modes. This applies to scalar UDFs written in Python, Java, Scala, JavaScript, and SQL. For incremental refresh, VOLATILE UDFs are supported only in the SELECT clause.

...

Read more

Snowflake Cortex AI_TRANSCRIBE now supports AAC (Advanced Audio Coding) input, both as a standalone audio file and as the audio track inside a supported video file. Because AAC is one of the most widely used audio codecs across podcasts, streaming, and video pipelines, you can now transcribe a wider range of real-world recordings directly in Snowflake, without first converting them to another format.

...

Read more

Select Integrated SaaS listings now let you start a free trial of a provider's application directly from a Marketplace listing. This experience replaces the trials previously offered through Snowflake Partner Connect and brings these trials into the Marketplace, where discovery, trial, and purchase all happen in one place. Anyone can browse and discover these listings in the Marketplace (no elevated privileges required), and you can return to them at any time to start a trial or purchase the application, including drawing down on your existing Snowflake capacity commitment through Marketplace Capacity Drawdown.

...

Read more

You can now use SQL to register, rotate, modify, and remove named key pairs for a user. Each key pair has its own name and supports an optional role restriction and expiration time.

...

Read more

Consumer-controlled maintenance policies for Snowflake Native Apps are now generally available.

...

Read more

CREATE OR ALTER is now generally available for 24 object types. CREATE OR ALTER creates an object if it doesn't exist, or modifies an existing object to match the statement definition. It's especially useful in declarative deployment workflows and CI/CD pipelines where the same script must be safe to re-run.

...

Read more

You can now clone a dynamic Iceberg table to a new dynamic Iceberg table or to a Snowflake-managed Iceberg table.

...

Read more

The Snowflake CoWork mobile app for iOS is now generally available on the App Store.

...

Read more

Cloud capacity isn't always guaranteed. When a compute pool's instance type is unavailable, provisioning fails with an insufficient capacity error (ICE), leaving the pool unable to scale. Previously, resolving an ICE required manually switching the pool to a different instance type. Backup instance types remove that manual step: you specify an ordered list of fallback instance families, and Snowflake automatically retries provisioning using the backup families in the order you specify.

...

Read more

Online Feature Store is now available in public preview. Serve the latest feature values with low-latency online retrieval backed by Snowflake Postgres for real-time ML applications such as fraud detection, recommendations, and personalization. Use batch, stream, and real-time feature views; ingest events through REST APIs; and integrate with real-time model inference.

...

Read more

• Per-user quotas (Preview)
• Semantic view materializations (Public preview)

The following updates are now available in Snowflake Data Clean Rooms:

• General performance improvements and bug fixes.
• Updates to private preview features.

Deep Research in Snowflake CoWork is now generally available. Deep Research is an investigation mode for complex, open-ended questions that require multi-step reasoning across your data. Instead of returning a single result, Snowflake CoWork decomposes the question into multiple sub-investigations, runs them in parallel across your structured and unstructured data, and synthesizes the findings into a structured report. Every claim in the report is traced back to its source data and queries, so you can verify the reasoning and share results with confidence.

...

Read more

Snowflake Cortex AI_TRANSLATE now supports translation of up to 100,000 input tokens, enabling organizations to translate large documents and long-form text directly in Snowflake. Customers can now translate documents, lengthy transcripts, and conversation logs that previously exceeded the function's context limits without manually splitting text or building custom preprocessing pipelines.

...

Read more

With this release, we are pleased to announce the general availability of domain verification.

...

Read more

Snowflake now supports resource budgets to control credit spending for Cortex Search services. Resource budgets let you define a monthly spending limit and take automated actions, such as revoking access or suspending the service, when spending exceeds your limits.

...

Read more

Snowflake Container Runtime CPU and GPU version 2.7 is now available. This release includes the following changes:

• Added support for ipywidgets in notebooks.
• The PyTorch backend now defaults to Ray Train. To opt out, set PYTORCH_USE_LEGACY_TRAINER=1. The legacy path will be removed in an upcoming release.
• TunerConfig now defaults resource_per_trial to {"CPU": 1} when uses_snowflake_trainer=True.

...

Read more

The following new features and enhancements are now available in Snowflake Data Clean Rooms:

• Manage data offerings and templates in Snowsight - Public Preview: You can now browse, search, inspect, and create data offerings and templates directly in the Snowsight UI.
• Drop/Leave support for failed collaborations: Collaborations with a status of CREATE_FAILED, CREATE_TIMED_OUT, or JOIN_FAILED can now be dropped or left, in order to clean up collaborations in your account.
• New region availability: Snowflake Data Clean Rooms is now available in the following regions: GCP Melbourne, Azure US-East-1, AWS Thailand (Bangkok), AWS Malaysia (Kuala Lumpur), and AWS New Zealand (Auckland).

With this release, Snowflake supports the VARIANT data type on Iceberg tables created from Delta table files in object storage (Delta Direct).

...

Read more

Snowflake is rolling out metadata-based sensitive data classification recommendations in Snowsight to help you identify databases that are likely to contain sensitive data and enable classification with one click.

...

Read more

Snowflake can now act as the OpenID Connect (OIDC) provider when Snowflake workloads authenticate to external services. To use this capability, create a secret of type WORKLOAD_IDENTITY_FEDERATION and share its issuer URL and subject identifier with the external service to establish trust. When the workload needs to authenticate, call SYSTEM$ISSUE_WORKLOAD_IDENTITY_FEDERATION_TOKEN to obtain a short-lived ID token to send to the service.

...

Read more

Analytical search, an orchestration capability in Cortex Agents that enables analytical queries over large document collections, is now in Public Preview.

...

Read more

Snowflake now supports replication of dynamic Iceberg tables as part of replication and failover groups. Previously, dynamic Iceberg tables were skipped during replication refresh operations.

...

Read more

You can now use an SQL query (rather than a physical table) as a logical table in a semantic view. This lets you define logical tables based on joins, transformations, or any SQL expression without creating a separate view first.

...

Read more

You can now define variables in a semantic view to parameterize facts, dimensions, and metrics. Variables let you customize calculations at query time without changing the view definition.

...

Read more

The following new features and enhancements are now available in Snowflake Data Clean Rooms:

• Edit Collaboration - Public Preview: The new Edit Collaboration API lets collaboration owners add or remove collaborators and change the roles of existing collaborators. Each requested change takes effect only after the affected collaborators approve it. This capability is available only for collaborations created after this update. For more details, see the API documentation.
• Improved API auto-upgrades: The retry mechanism for auto-upgrade is improved for better task efficiency. Please enable auto-upgrades to always be on latest version. Details here. Please remount if you are on version below 12.3. Learn more here.
• Updated details column for update requests: The details column payload returned by VIEW_UPDATE_REQUESTS has been updated for several update request types, with some fields removed or renamed for greater clarity and consistency. The following fields have been removed from the details payload:

...

Read more

When defining a semantic view, you can now specify sample values and enum indicators in the create semantic view command.

...

Read more