Control Notifications by View

You can now mute weekly and monthly trend notifications for individual views, so shared notification channels aren't flooded with routine cost updates that bury the anomaly alerts requiring action. The Global View now controls out-of-the-box anomaly detection and can suppress every global anomaly, including system-generated alerts prefixed "Global," while each view can independently send weekly notifications, monthly notifications, both, or neither.

CloudZero MCP

Organization admins can now provision, rotate, and revoke static OAuth clients through the API, so enterprise agent platforms that require pre-configured credentials can connect without dynamic registration.

AI Signals Default Dimensions

New organizations are automatically provisioned with ten default grouping dimensions for AI cost analysis, including Model Provider, Team, Client, and GitHub Repo.

Monthly Normalization in Explorer

A Normalized trend button now appears in the chart controls when monthly granularity is active. A dashed trend line overlays the 30-day-normalized value on each bar, and the tooltip displays both actual cost and the 30-day equivalent side by side. The setting persists in the URL, so normalized views survive a refresh and can be shared directly with teammates.

Improved

- Optimize: Recommendation descriptions now cite the cost figure matching the recommendation's own cost type, removing inconsistencies across billing perspectives.

  • Explorer: The chart card is now shorter so that both the chart and the cost table are visible together without scrolling.
  • UX: The 'Classic Experience' shortcut has been removed from the sidebar as the platform completes its transition to the new experience.
  • Optimize: The ServiceNow Update Set download has been removed from the integration setup to prevent duplicate records on re-import.
  • UX: The command palette (Cmd+K / Ctrl+K) now features a compact search dock with a dedicated Shortcuts tab, saved views, direct link to Settings, and search across Platform pages.
  • Telemetry Streams: New AI telemetry sources can now be onboarded without waiting for a dedicated deployment, reducing the time to start ingesting data from a new provider.
  • AI Signals: All context convention attributes, including workflow name, user roles, and version control details, are now fully recognized during ingestion so your telemetry data maps to the correct dimensions.
  • Optimize: Rightsizing recommendation titles now highlight only the attributes that changed, such as IOPS or throughput, making it easier to identify what each recommendation adjusts.
  • Optimize: Kubernetes rightsizing recommendations now scope the cost graph to the specific workload's cluster, namespace, and pod, so you see that workload's actual cost.
  • AI Signals: My Work now displays a coverage indicator showing what percentage of your AI spend is labeled, so you can track categorization progress over time.
  • AI Signals: Internal pipeline dimensions that were inadvertently visible to customers are now hidden, so only the ten customer-facing grouping dimensions (AI User, AI Model, etc.) appear.
Fixed

- Analytics: Embedded dashboards no longer reuse a previous user's session or show a stale login page when two users log in on the same browser, ensuring each user always sees only their own data.

  • Datadog Connection: Re-ingesting a Datadog billing month older than one month no longer drops the daily cost distribution, preserving the accurate day-by-day spend shape.
  • Optimize: Customers with large Jira project lists can now load the integration settings and create work items without encountering a timeout.
  • Optimize: EKS upgrade recommendations now reflect current Kubernetes version support and accurate cost figures instead of outdated guidance.
  • Optimize: The View in Explorer link on Kubernetes recommendations now carries the workload filter and correct grouping, so the Explorer view matches the recommendation's cost chart.
  • UX: Pages no longer hang on an infinite loading skeleton after switching organizations; all routes reload correctly without requiring a manual page refresh.
  • Budgets: CSV import errors now display the specific reason the file was rejected, such as invalid columns or rows, so you can correct and re-upload without guessing.
  • Optimize: Pod details now appear in the Kubernetes recommendation detail view, where a field-name mismatch previously prevented them from displaying.
  • CloudZero MCP: Sessions now automatically re-authenticate when you switch organizations, ensuring cost queries always return data for your current active organization.
  • macOS Collector: Upgrades deployed through MDM no longer time out when the collector is already running but its background agent is unregistered, so managed installs complete reliably.
  • macOS Collector: Proxy handling on MDM-managed machines is now more reliable — the collector correctly selects the MDM-delivered configuration, disables redundant settings so administrator ordering priority is preserved, and resets and recovers on its own after a profile is removed.
  • AWS Connection: Connections now retain the name you entered during setup instead of replacing it with a generated label on certain upgrade paths.
  • Explorer: The "View in Explorer" link from insights no longer hangs on a blank page when the URL contains a custom date range without valid dates; Explorer now normalizes bare dates or falls back to the default 30-day range.