Configuring Explorer filters requires knowing how your data is structured — which dimensions exist, what values they contain, and how to combine them to isolate the costs you care about. Filter Assist removes that prerequisite. Describe what you want to see in plain language and CloudZero translates it into a working filter configuration automatically.

Ask for something like "show me untagged EC2 costs in production" or "Kubernetes spend by team excluding the platform namespace" and Filter Assist builds the filters for you. Every result is fully visible and editable — nothing is a black box. If the suggestion isn't quite right, adjust it directly in the filter bar the same way you would any manually configured filter. Access it via the sparkle icon in the Explorer filter bar.

RBAC for CostFormation Namespaces

Namespace-level access control is now generally available — restrict who can view and edit Dimension definitions per namespace, delegate ownership to specific teams, and run multi-team CostFormation strategies without exposing every definition across the org. Available in the app, Public API, and VS Code Toolkit. See Namespace Access Control.

New Features

AWS Parquet CUR Support: CloudZero now ingests AWS Cost and Usage Reports in Parquet format — connect your existing Parquet CUR directly with no additional setup.
Organization Settings: Org admins can now rename the organization display name, broadcast announcements to all users, and set whether Explorer cost trends are shown or hidden by default across the entire org. Available under Settings > Organization Settings.
Kubernetes Integrations: You can now hide and unhide individual clusters from the K8s integration page, keeping your workspace focused on relevant infrastructure.
Connections: When enabling a new connection, you can now specify how far back to ingest historical cost data, giving you control over the initial data window.
Anthropic Connector: Anthropic Enterprise is now supported as a cost data source, with up to a year of historical usage ingested on first connect.
Optimize: Recommendations can now be sorted by two fields simultaneously, making it easier to organize and prioritize cost-saving opportunities.
Explorer: A "Year to Date" preset is now available in the timeframe selector, instantly showing spend from January 1 through yesterday.
Explorer and Views: New "is null" and "is not null" filter operators let you isolate untagged or unallocated costs that previously had no way to be filtered.
User Settings: A chart color palette selector is now available in the Preferences tab, letting you persist chart colors across the platform.
Explorer: Query History and Features are now unified in a single Configurations side panel with History and Features tabs.

Improvements

Connections: Connections now show a degraded (yellow) status when partial data is received, so you can immediately tell when a Snowflake share is missing tables rather than seeing a healthy green.
Anthropic Connector: Cost data is now attributed to your Anthropic organization ID and user ID, giving you more accurate account-level and user-level cost breakdowns.
Explorer: Hourly granularity is now available for date ranges up to 35 days (up from 30), allowing full calendar months to use the hourly view.
Explorer: Previous and next arrow buttons flanking the date range picker let you shift the current time window forward or backward in one click.
Explorer: The comparison date picker has been redesigned with a wider layout, preset shortcuts, and always-visible dual calendars, making period-over-period comparisons easier to configure.
Optimize: Filters now support "contains" and "does not contain" operators, and Select All respects your current search, making it faster to build precise recommendation filters.
Optimize: The recommendation detail panel now includes editable Status and Effort dropdowns so you can update recommendation status inline without leaving the overview.
Explorer: "Idle" and "(unallocated)" are now available as filter options for Kubernetes and custom dimensions, restoring parity with legacy behavior for identifying unassigned costs.
Dimension Studio: The History tab now shows the email of the person who last edited a namespace, making it easier to track who made recent changes.
Analytics Dashboards: Conditional dashboard visibility now covers 18+ providers including Snowflake, Databricks, Datadog, and OpenAI, so the right dashboards appear automatically for your connected services.
K8s Integrations: Labels, annotations, and patterns for a Kubernetes agent now open in a full-height side panel instead of a small popover, making long lists much easier to read and search.
Help button: The Help and Guidance button is now more visible with a high-contrast orange style and can be dragged to any position on screen, with position remembered between sessions.
Budgets: A "View in Explorer" button is now available on the budget detail screen, letting you jump directly to the cost breakdown behind a budget.
CSV Upload: You can now target dimensions that are still publishing when doing a CSV upload, closing a parity gap with the classic experience.

Fixes

Azure Connections: Full unload jobs no longer leave orphan files when retried, ensuring cost data is complete and not duplicated across billing periods.
Explorer: Cost growth from zero now shows a "New" label instead of the confusing infinity symbol, making spend comparisons easier to read.
Connections: Partition deduplication now checks against raw ingested data instead of cursor state, preventing missed or duplicate cost records during incremental syncs.
Anthropic Connector: Web search costs in the Anthropic cost view are now calculated correctly and no longer inflated by the number of models in use.
Optimize: Database Savings Plan recommendations now show one accurate recommendation per payer account instead of duplicates.
Telemetry Streams: Last Request and Start Date timestamps now display the correct UTC times instead of showing values shifted by the browser timezone offset.
Billing Data: Cost and usage export rows are now sorted deterministically, preventing downstream consumers from seeing data appear missing between runs.
Explorer: Grouped cost queries no longer fire with missing date parameters, preventing blank or incorrect cost data from appearing in filtered views.
Filter Assist: Filters with multiple operators on the same dimension are now preserved correctly instead of silently dropping the second operator.
Dimension Studio: Split transform index now defaults to 1 (matching the backend 1-based indexing), eliminating off-by-one confusion when splitting dimension values.
Explorer: Bookmarked or shared Explorer URLs referencing deleted dimensions now recover gracefully to a usable view instead of showing an error.
Explorer: Legacy bookmarked URLs with date ranges like "Last 30 Days" now open with the correct time range applied instead of being ignored.
Settings: Organization preferences now save reliably in all environments, fixing a failure where the update could silently fail due to a missing record.
Dimension Studio: Dimension rules with a single source no longer cause a crash when the Sources field contains a plain string value instead of a list.
Dimension Studio: Multi-word dimension names with spaces now return correct search results.
Dimension Studio: Search now finds dimensions by their display name in addition to the raw key, so dimensions are no longer missing from search results.
Telemetry Streams: Recreating a stream with the same name now shows current data immediately instead of displaying stale details from the deleted stream.
Natural Language Search: Filter suggestions no longer show duplicate entries when querying by service name.
API Keys: The Last Modified column now sorts by date correctly instead of alphabetically, and defaults to showing the most recently modified key first.
Views: Filters and group-by state are now carried over when creating a view from Explorer, so your current configuration is no longer lost.