What's new in GitLab 18.6

Nov 20, 2025
Past release

From configuration to control: Enhanced security and productivity through modern interfaces and brand new platform capabilities to keep your software teams in flow. Note: After this page was originally published, the default Security Manager role was withdrawn from the release. It will be included in a future update. The content below is updated for accuracy.

Enterprise governance meets development needs

GitLab 18.6 enables teams to move fast within clear boundaries, building software their way — with their standards and their approved technologies.

Read CEO's blog

Self-managed instances now default to the streamlined interface that keeps teams in flow. With this update, teams can:

  • Minimize context switching with side-by-side panels that eliminate tab-juggling during debugging.

  • Maximize code visibility with dynamic layouts that adapt to your current task, whether you’re reviewing, coding, or planning.

  • Experience the same consistent interface and unified navigation across GitLab.com and self-managed deployments.

  • Retain flexibility through feature flags for teams requiring gradual transitions.

Find any pattern across millions of lines instantly — enhanced by AI-powered suggestions when enabled. With exact search, developers can:

  • Locate code in seconds using regex patterns across entire instances without local clones.

  • Start immediately on GitLab.com with zero-configuration deployment already active.

  • Deploy in secure environments where Zoekt runs completely offline for air-gapped networks.

  • Combine with AI assistance (when enabled) for intelligent search suggestions based on natural language.

Full IDE capabilities now available in air-gapped environments, including optional AI assistance for classified networks. Users can:

  • Remove all external dependencies by hosting Web IDE assets internally without internet connectivity.

  • Enable full development features including markdown preview, extensions, and debugging in isolated networks.

  • Meet strict compliance requirements for government, defense, and financial institutions.

Apply business context to security data — enhanced by AI that learns your prioritization patterns. With security attributes, teams can:

  • Reduce triage time by filtering "Production + Internet-facing" vulnerabilities first.

  • Filter by business unit tags to help identify which teams should handle specific vulnerabilities.

  • Bulk dismiss false positives with one click, removing test file noise from critical paths.

  • Let AI suggest priorities (when enabled) based on your historical remediation patterns.

Components now know their own metadata — from commit SHAs to version numbers — to eliminate version drift, with AI assistance for configuration when enabled. Users can:

  • Eliminate version mismatches with spec:component ensuring Docker images match component versions automatically.

  • Reduce pipeline complexity using dynamic dependencies that adapt to your matrix builds.

  • Prevent "works locally" failures by ensuring consistent versioning across environments.

  • Get AI configuration help (when enabled) for complex multi-platform scenarios.

Helm charts without limits

Deploy

Manage

Operate

Unlimited chart storage with pre-computed metadata for instant access to any version. Platform teams can:

  • Access any historical version without 404 errors disrupting critical rollbacks.

  • Improve performance with pre-computed metadata eliminating on-demand generation.

  • Centralize chart governance without splitting across repositories for scale workarounds.

  • Maintain deployment history across all releases for simpler compliance and debugging.

Simplified ownership management that respects organizational hierarchy and is enhanced by AI review when configured. Teams can:

  • Eliminate redundant configuration with parent group permissions automatically applying to projects.

  • Reduce approval delays as the right reviewers already have access through inheritance.

  • Keep existing CODEOWNERS files working without migration or changes required.

  • Add AI pre-review (when enabled) before human code owners for faster feedback cycles.

GitLab-built agents for planning and security are now available by default in the Agentic Chat dropdown. When activated, teams can:

  • Break down complex work instantly with the GitLab Duo Planner Agent, turning vague requirements into implementable tasks.

  • Maintain milestone alignment as the Duo Planner Agent respects your sprint deadlines automatically.

  • Understand vulnerabilities quickly with the GitLab Duo Security Analyst Agent that explains CVEs in plain language.

  • Access both agents on GitLab.com and self-managed instances.

Cloud-connected self-managed users can now select which AI model powers their GitLab Duo Agentic Chat experience directly in the GitLab UI. This gives teams more flexibility to manage performance, cost, and governance while choosing models that best fit their environment — whether on-premises, in a private cloud, or in a public cloud. This gives teams:

  • Greater control over model performance, accuracy, and cost.

  • The ability to choose regionally compliant or in-house models for data residency needs.

  • Flexibility to match models to organizational policies and priorities.

  • Consistent model selection across any self-managed deployment environment.