By Erez Schnaider, Technical Product Marketing Manager, ClearML
ClearML Enterprise v3.27 delivers on the three capabilities most requested by practitioners : clear visibility into compute consumption inside projects, simpler and safer access control for remote sessions and deployed endpoints, and quality-of-life upgrades across the UI. The result is better cost control, stronger governance, and faster day-to-day execution.
New Project Workloads Dashboard
As teams grow, understanding who uses which resources and when becomes critical for billing & chargebacks, budgeting, and optimization. The new Project Workloads Dashboard summarizes a project’s compute usage with breakdowns by resource type, user, and sub-project. It provides daily trends and period totals so you can quickly see top consumers and spot anomalies.
What’s new
- View utilization by resource, user, or sub-project, then drill into spikes to investigate tasks behind them.
- Track usage over time to validate policy changes or autoscaling settings.
- Use the insights to inform quotas, reservations, and scheduling priorities.
Access it from any project’s Workloads tab.
Enhanced AI Application Gateway token management
The ClearML AI Application Gateway provides secure, authenticated access to remote development sessions and deployed model endpoints in enterprise environments. In v3.27, token management is more transparent and easier to administer. You can list all active tokens, add labels for quick identification, and revoke tokens instantly. These improvements help administrators supervise access at scale while keeping developer workflows smooth.
What’s new
- Centralized view of active tokens
- Labels to organize tokens by user, purpose, or environment
- One-click revocation to remove access immediately
New global search, clone naming, and data workflow wins
v3.27 adds an advanced mode for global search so power users can specify API-style filters directly when they need precise queries. You can also customize the default name format of cloned tasks, which helps keep large projects tidy. For data-centric workflows, you can delete Hyper-Datasets directly in the UI and open the Hyper-Dataset Frame Viewer straight from Dataviews to cut down on navigation clicks. Single-value scalar comparisons are clearer too, since each variant gets its own plot when grouping by metric.
What’s new: quality-of-life improvements
- Delete Hyper-Datasets in the UI
- Better single-scalar comparisons for grouped metrics
- Open Frame Viewer directly from Dataviews
- Custom default naming for cloned tasks
- Global Search advanced mode for precise filtering
New SSH credential vaults for remote sessions
Administrators can now manage SSH server credentials in dedicated vaults to streamline and secure access to remote sessions such as the SSH Session app or ClearML Session sessions. Centralizing credentials simplifies provisioning while improving governance.
More reliability and polish across the platform
v3.27 includes dozens of bug fixes across tasks, pipelines, reports, object table filters, and the UI. Highlights include improved handling of large console logs, more accurate task status updates in tables, better tag behavior, and multiple fixes in the Hyper-Dataset frame viewer and Global Search. Teams should see smoother comparisons, cleaner tables, and fewer interruptions during investigations.
Who benefits
- AI Builders get visibility to manage costs and capacity, plus stronger token governance.
- Data scientists and engineers move faster with simpler dataset cleanup, clearer comparisons, and quicker hops into frame-level inspection.
- Admins and security gain central controls for SSH credentials and tokens, with immediate revocation when needed.
How to get started
- Open any project and select the Workloads tab to explore usage by resource, user, or sub-project.
- Visit Application Gateway settings to review tokens, add labels, or revoke access. The App Gateway is available with ClearML Enterprise.
Read the full v3.27 release notes to see every improvement and fix.