Bin and group

You can now turn numeric dimensions into ranges or collect string values into named groups without changing the data model. Open a dimension's menu in the Data Pane, choose Create bins or Group values, and Cube adds a query-scoped field you can use immediately.

Define numeric ranges manually or by width, choose how their labels appear, and keep values outside the range instead of dropping them. For strings, search values, name each group, and decide where everything else belongs. You can reopen the field later and edit it in place.

Other improvements and fixes:

Workbooks

Sort and row limit now share one Sort & limit popover, keeping both controls together while leaving more room in the report toolbar.

Visualization

• Table charts now support collapsible row groups, including per-level default states and expand-all or collapse-all controls
• The chart picker now recommends one best-fit visualization as your query changes and can apply a high-confidence choice to a new report
• Bar charts now use Grouped, Stacked and Stacked % entries with a separate Orientation control, so switching between vertical and horizontal preserves formatting
• Donut charts can now show the measure total in the center

Administration

• Billing now has separate Overview and Cost and Usage pages, with searchable paginated tables and clearer loading and error states
• Alerts now live in the Admin navigation alongside Billing and Audit Log
• Fixed Generated SQL in Query History showing bare ? placeholders instead of filter values

AI

• The Auto agent now uses Claude Sonnet 5, with Claude Opus 5 for data modeling on paid plans
• Analytics Chat can now run a dbt pull end to end, follow its progress, open the generated branch and summarize new cubes
• Fixed MCP connectors repeatedly asking you to reconnect when their OAuth access tokens expire
• API Credentials now includes a Chat API tab with the deployment URL, a curl example and Cube authentication guidance

Integrations

• Fixed default UI filters from a semantic view being ignored when you start an exploration in the Google Sheets and Excel add-ons
• Fixed the Google Sheets and Excel add-on side panel not scrolling when its tabs contain more controls than fit in the window

Modeling

• The model editor now recognizes the grain parameter for multi-stage measures, including autocomplete for its fields
• Fixed Metadata API table lineage omitting source tables for cubes defined with sql, including joins, CTEs and derived tables

API

• Personal Core Data API tokens now authenticate REST requests as well as SQL
• Cube CLI now supports cube validate against production, a branch or the active dev-mode model, returning compiler errors and a non-zero exit code for CI
• Cube CLI now supports cube spec to search or export the live OpenAPI specification, while deployment settings are now available through the public REST API
• Cube CLI now supports cube data-model enable-branch and disable-branch for keeping shared-branch staging environments active from CI; the public REST API can also delete data-model branches
• Fixed Tesseract treating literal % and _ values as wildcards in contains and notContains filters, including queries served from pre-aggregations

Workspace

• The Last viewed by you cell now shows when a Workbook or report was created and last edited, and opening it no longer changes that edit date
• Copy link is now available inside share dialogs and on Workbook and folder rows, with clear copied-state feedback