This release strengthens enterprise-grade AI governance with platform-wide multi-factor authentication (MFA), deployment-wide directory sync, and more granular project-level access controls. We’ve expanded observability with fully multimodal traces, images, audio, and PDFs now render inline, giving teams richer visibility into model and agent behavior. Governance and framework pages now surface rule-result summaries to streamline audit readiness and compliance reporting. We’ve also added new integrations, SDK enhancements, API controls, and CI-compatible CLI improvements to support automation across modern AI workflows.
Features
•SecurityMulti-factor authentication (MFA)
•UI/UXImages and audio render inline in trace steps and data tables
•UI/UXPDF preview in traces and data tables
•SecurityDeployment-wide directory sync with project-level access group support
•UI/UXGovernance and frameworks pages now show a summary of rule results
•PlatformData download in development mode
•On-PremDiagnostics page for testing connections and sending test emails
•IntegrationsLangflow integration
•SDKsOpenAI Responses API support in TypeScript SDK
•APIPull rows from specific test results via API
•APIDelete projects via API
•PlatformCustom metrics bundle now viewable and downloadable from the UI
•IntegrationsMCP: custom functions callable from Openlayer
Improvements
•UI/UXOpenlayer system tags renamed to "categories"; user-defined labels remain "tags"
•PlatformStatus page now reflects real-time system health
•CLIopenlayer push wait mode is now CI-mode compatible
•CLIopenlayer metrics run now supports a -force flag
•CLIopenlayer metrics run prints which metrics are being run
•PlatformProject owners now have admin-level access for all project-level actions
•SDKsSession and user IDs now accept a broader range of characters
•SDKsLangChain tracer now supports VertexAI API Keys
•SecurityDirectory sync now supports a NoAccess group level for finer-grained access control
•SDKsConversationalSearchService traces now include document chunks and nested step hierarchy
•UI/UXTest overview column shows more detail at a glance
•UI/UX"Advanced settings" renamed to "Evaluation settings" in test configuration
•UI/UXImprovements to the manual evaluation window config in test creation
•APIAPI key authentication now supported for test result endpoints
Fixes
•SDKsLangChain callback handler now correctly attributes sessions and user IDs
•UI/UXCustom metrics no longer show a duplicate tag in the test creation page
•PlatformEmail notifications no longer sent to users without access-group access to a project
•On-PremCopy buttons now work correctly in on-prem deployments
•PlatformOTel traces no longer record empty inputs and outputs
•UI/UXContext relevancy test now correctly indicates when ground truth is being used
•PlatformRagas metrics no longer fail intermittently with key errors
•PlatformPrompt injection tests now return consistent row counts
•APIData export now correctly applies the selected start and end time range
•PlatformProject-level environment variables now apply correctly to commits
•UI/UXDetected PII types now render as tags in test results
•IntegrationsFixed a constraint that prevented the same Copilot Studio bot config across multiple workspaces
•UI/UXTrace data no longer overflows the trace detail modal
$ openlayer push
Stop guessing.
Ship with confidence.
The AI governance and observability platform
We value your privacy
We use cookies to enhance your browsing experience, serve personalized content, and analyze our traffic.
