Updates 2025
December 22, 2025
Features & Enhancements
Set associated projects when creating datasets
Apply project associations at dataset creation time
Ensures datasets are accessible to intended projects from the start
Support runtime config preset slug in step config
Allow referencing presets by slug in YAML configuration
Provides more flexible preset selection
Use step preset in execution creation UI
Automatically apply preset from step definition in execution forms
Ensures UI respects YAML-configured presets
UI/UX Improvements
Improve logging for organization member addition
Add detailed logging when adding members to organizations
Aids troubleshooting and auditing of access changes
Bug Fixes
Do not assign plans from disabled organizations
Skip plan assignment from organization memberships if organization is disabled
Prevents access grant through inactive organizations
Fix image comparison layer blending
Refactor blending algorithm for more accurate image comparisons
Ensures color differences are displayed correctly
December 17, 2025
Features & Enhancements
Support Warden machine info for all VM environments
Extend Warden machine listing to all VM-based environment types
Provides consistent machine inventory across platforms
Allow opting out of counting executions
Add option to skip execution counting in execution table views
Improves performance for very large execution lists
Don't expose inaccessible datums when filtering via public store
Enforce access control when querying data through public stores
Prevents information leakage across organization boundaries
Get runtime config preset from step configuration
Read preset ID directly from step config when available
Ensures preset selection is consistent with YAML definitions
UI/UX Improvements
Improve notebook git credentials documentation
Add description clarifying notebook git credentials are personal
Helps users understand credential scope and security
Improve organization page descriptions
Standardize and refine description text across organization settings
Provides clearer guidance for configuration options
Bug Fixes
Fix confusion matrix long labels
Properly truncate and display long labels in confusion matrix visualizations
Prevents layout issues with verbose class names
Fix confusion matrix value formatting
Use precise values in tooltips and concise notation in cells
Balances readability with detailed information
Fix runtime config cache volumes on copy
Properly preserve cache volume configuration when copying executions
Ensures copied executions maintain caching behavior
Disable commit changes for adhoc execution copying
Prevent commit selection when copying adhoc executions
Avoids confusion since adhoc executions don't have associated commits
December 10, 2025
Features & Enhancements
Add suggest_name to deployment version creation
Provide automatic version name suggestions based on content
Speeds up deployment version creation with sensible defaults
Use cache volumes from step YAML
Read cache volume configuration from valohai.yaml step definitions
Enables declarative cache volume setup
UI/UX Improvements
Allow pasting dataset URL as deployment file
Accept dataset version URLs directly in deployment file input
Simplifies linking datasets to deployments
Markdownify execution parameter descriptions
Render parameter descriptions with markdown formatting
Supports richer documentation in execution forms
Bug Fixes
Properly serialize parameter definitions
Fix serialization of parameter metadata in API responses
Ensures parameter information is complete and correct
Hide disabled organizations
Filter disabled organizations from organization listings
Prevents confusion from inactive accounts
Use id lookups instead of ctime where possible
Replace timestamp-based lookups with ID-based queries
Improves performance and reliability of database queries
December 3, 2025
Features & Enhancements
Add Oracle Object Storage wrapper
Support Oracle Cloud Infrastructure Object Storage as store backend
Expands storage options for data and models
Mount cache volumes on workload container
Attach cache volumes to the main workload container
Improves cache accessibility for user code
UI/UX Improvements
Update runtime values on commit change
Refresh runtime configuration when switching commits
Ensures execution settings stay synchronized with code
Bug Fixes
Import dataset files only once
Prevent duplicate file imports when creating datasets
Ensures dataset versions don't contain redundant entries
Fix serializing store with experimental URL signer
Handle store serialization when using experimental presigned URL feature
Prevents API errors with advanced storage configurations
December 1, 2025
Features & Enhancements
Implement ExecutionHashV2
Introduce improved execution hashing algorithm for better deduplication
Provides more accurate identification of duplicate executions
Add property keys overhaul
Improve metadata property key handling and performance
Enhances reliability of metadata queries
UI/UX Improvements
Enhance create execution and task view layouts
Streamline layout and organization of execution creation forms
Makes it easier to configure and launch executions
Bug Fixes
Ensure stores patched don't mess with paths
Prevent path corruption when applying store configuration patches
Ensures data remains accessible after store updates
Improve datum adoption error handling
Treat pre-existing datums as success rather than failure
Reduces false error reports during data import
Do not allow concurrent automatic refreshes in Tablor
Prevent race conditions from overlapping table refresh operations
Improves stability of data tables
Fix hidden commit usage in copied configurations
Properly set configurations from hidden commits as default
Ensures copied executions work correctly
Remove user usage caching
Eliminate stale cache issues with user quota information
Ensures usage data is always current
November 25, 2025
Features & Enhancements
Allow selecting previously used datums for data inputs
Show historical datum selections for workflow inputs
Speeds up workflow creation by reusing previous inputs
Override step inputs with pipeline node definitions
Allow pipeline node configuration to override step defaults
Provides more flexibility in pipeline parameterization
UI/UX Improvements
Lazy load project tag colors
Load tag colors on-demand to improve initial page load
Reduces data transfer for projects with many tags
Upgrade organization email logo to new brand colors
Update email templates with current brand styling
Ensures consistent visual identity
Improve visual separation in file upload UI
Distinguish between previously used and newly uploaded files
Makes file selection clearer in workflow inputs
Bug Fixes
Merge cumulative metadata instead of overwriting
Preserve existing cumulative metadata when adding new values
Prevents loss of metadata from partial updates
Optimize Bayesian tasks' execution reuse
Improve query performance for Bayesian optimization workflows
Speeds up task execution with reuse enabled
Fix N+1 queries in notification emission
Reduce database queries when sending notifications
Improves performance of notification system
November 19, 2025
Features & Enhancements
Rename workflow definitions to blueprints
Update terminology from "workflow definition" to "workflow blueprint"
Provides clearer naming for workflow templates
Allow setting CPU max for deployment endpoints
Support configuring CPU limits for Kubernetes deployment endpoints
Gives control over resource allocation for serving infrastructure
Allow token policies only for organization tokens
Restrict token policy feature to organization-level tokens
Simplifies security model and reduces confusion
Support dataset versions for deployment files
Accept dataset version URLs when creating deployment endpoint files
Simplifies deployment version creation
UI/UX Improvements
Apply dataset tag colors in project data view
Show colored tags in dataset listings
Improves visual organization and scanning
Autoselect latest commit and first blueprint
Pre-select sensible defaults in workflow creation form
Reduces clicks needed to create workflows
Allow editing store configuration JSON in admin
Enable direct JSON editing of store configuration for advanced users
Provides escape hatch for complex configurations
Bug Fixes
Fix social account disconnect
Resolve issues when disconnecting OAuth authentication providers
Ensures users can manage their connected accounts
Do not allow default values for required assigned inputs
Prevent setting defaults on inputs that must be explicitly provided
Ensures required inputs are always reviewed
Exclude deleted projects from team details API
Filter out soft-deleted projects from team project lists
Prevents confusion from stale project references
Tell Scalie to forget disabled environments
Clean up scaling service state when environments are disabled
Prevents resource leaks in autoscaling system
November 10, 2025
Features & Enhancements
Add Warden machine listing
Provide API for listing available machines from Warden service
Enables dynamic machine inventory for VM-based environments
UI/UX Improvements
Show runtime config preset editor for Kubernetes environments
Allow configuring runtime presets for Kubernetes execution environments
Makes preset management available for all environment types
Optimize SME workflow catalog
Only fetch displayed workflow run history instead of all runs
Improves loading performance for workflow catalog
Bug Fixes
Fix double-POST with unprocessed inputs
Prevent duplicate workflow creation requests when inputs are being uploaded
Ensures only one workflow run is created
Set and delegate input defaults properly
Correctly apply default values to workflow inputs
Ensures workflows receive expected default configurations
Fix error messages in pipeline editor
Improve error handling for missing commits and library steps
Provides clearer guidance when configuration is incomplete
Fix execution step icons
Display correct icons for different step types
Ensures visual consistency in execution views
November 5, 2025
Features & Enhancements
Allow key existence checks in property queries
Support checking if metadata keys exist using special syntax
Enables filtering data based on presence of properties
Add basic runtime options for Kubernetes workers
Allow configuring CPU limits and other runtime settings for worker pods
Provides control over resource allocation
Add custom execution iframe tabs
Support embedding custom tools and visualizations in execution views
Enables integration with third-party analysis tools
Support dataset versions for deployment files
Allow using dataset version URLs when creating deployment files
Simplifies linking models to specific dataset versions
Paginated and lazy organization tags
Load organization tags on-demand with pagination
Improves performance for organizations with many tags
UI/UX Improvements
Show execution input clear button for long lists
Add button to clear all inputs when many are present
Speeds up configuration when starting fresh
Rename image comparison views
Update view names to be more descriptive (e.g., "View 4" to "Split")
Makes it clearer what each comparison mode does
Fix image compare colorization with transparent images
Properly handle alpha channel when applying color overlays
Ensures transparency is preserved in comparisons
Add commit API filters
Support filtering commits by fetch_ref and commit_ref
Makes it easier to find specific commits via API
Bug Fixes
Show error if SME user lacks project store access
Display clear error when workflow inputs reference inaccessible storage
Prevents confusing failures during workflow execution
Validate endpoint image reference format
Check image references at API time and before packaging
Catches configuration errors earlier in deployment process
Sanitize weird file paths
Handle unusual characters and patterns in file paths gracefully
Prevents errors when adopting files with special characters
Fix DatumBrowser sticky preview and purged selection
Correct preview panel behavior and handle purged items
Ensures datum browser remains stable
October 22, 2025
Features & Enhancements
Add root path option when adopting files
Allow specifying a root path prefix when adopting files from storage
Simplifies organizing imported data with consistent directory structure
Apply tag renaming to existing tags
Automatically update all usages when renaming organization tags
Ensures tag consistency across all resources
Add interactive executions
Support creating executions with interactive mode flag
Enables debugging and development workflows
Virtualize adoption file list
Use virtual scrolling for large file lists during data import
Improves performance when adopting thousands of files
Add ANSI color support for console output
Render ANSI color codes in execution log viewer
Preserves colored output from tools and scripts
Add workflow run filter backend
Enable filtering workflow runs by various criteria via API
Makes it easier to find specific workflow executions
UI/UX Improvements
Show app version as logo tooltip for private installs
Display version information on hover over logo
Helps identify which version is deployed
Allow seeing deployment target details
Add view for examining deployment target configuration
Provides visibility into deployment infrastructure settings
Bug Fixes
Eradicate default exports in codebase
Replace default exports with named exports throughout frontend
Improves code maintainability and refactoring safety
Fix Tablor selection background color
Correct background color for selected rows in tables
Ensures selections are clearly visible
October 15, 2025
Features & Enhancements
Allow exposing node port services for Kubernetes Executions
Support exposing services via NodePort in Kubernetes
Enables external access to services running in executions
UI/UX Improvements
Add error handling to dataset readmes
Display graceful errors when readme content cannot be loaded
Prevents view crashes from missing or corrupted readme files
Bug Fixes
Fix long step names breaking layout
Prevent layout overflow when step names are very long
Ensures UI remains usable with lengthy configuration
Hide API call button from light users
Respect user capability flags when showing API documentation links
Prevents confusion for users without API access
Fix execution image comparison layout shift
Eliminate content jumping during image comparison loading
Provides smoother visual experience
Unset default environment groups for light users
Clear environment variable group selection for restricted users
Prevents access to environment configuration light users shouldn't modify
October 9, 2025
Features & Enhancements
Add dataset readme support
Allow attaching readme files to datasets and dataset versions
Provides documentation and context for datasets
Add model-hub-only user flag
Enable users with access restricted to model hub only
Supports specialized user roles focused on model management
Add associated projects to datasets
Link datasets to multiple projects for cross-project sharing
Enables dataset reuse across team boundaries
Add SME pipeline creation with inputs
Support configuring pipeline inputs in SME workflow creation
Provides more control over workflow execution
Add custom execution iframe tabs
Allow defining custom tabs in execution views via configuration
Enables embedding custom tools and visualizations
Copy step mounts into executions
Transfer mount configuration from step definitions to created executions
Ensures execution has access to all configured storage mounts
Automatically create pipeline deployments from YAML
Generate deployment configurations from valohai.yaml pipeline definitions
Simplifies deployment setup for pipelines
Add wildcard expansion limit configuration
Prevent excessive input expansion with configurable limit
Protects against accidental resource exhaustion from wildcards
UI/UX Improvements
Add "# Outputs" column to executions table
Show output count directly in execution listings
Makes it easier to identify executions with results
Improve grouped metadata plot
Add value filtering and grouping by metadata values
Brings pivot-table-like capabilities to metadata visualization
Improve zoom and pan in image comparison
Add expanded zoom area for execution image comparison
Makes it easier to examine fine details in output images
Add compact mode to execution image comparison
Provide condensed view option for comparing many images
Improves usability when comparing large image sets
Unify and fix React Select styles
Standardize appearance of dropdown components across the interface
Ensures consistent user experience
Show unified tags on executions, pipelines, tasks
Display tags consistently across resource types with visual limits
Makes tag-based organization more consistent
Bug Fixes
Fix project URL resolution with 'datasets' in name
Properly match URL fragments instead of substring matching
Resolves navigation errors for projects with reserved words in names
Prevent filter state leakage between datum browsers
Isolate filter state per view to prevent cross-contamination
Ensures filters don't carry over unexpectedly
Fix DatumBrowser sticky preview and purged selection
Correct preview panel behavior and selection state
Improves reliability of datum browsing interface
September 24, 2025
Features & Enhancements
Allow using single Ingress for whole Deployment Target
Support shared Kubernetes Ingress configuration across deployments
Simplifies network setup for multiple endpoints
Add datum sharing flag to stores
Control whether data in a store can be shared across projects
Provides finer-grained access control for stored data
Mount cache volumes on Kubernetes Executions
Add API for defining cache volumes in runtime configuration
Improves performance by caching data across execution runs
Add environment PVC proxy API
Provide API for managing persistent volume claims in Kubernetes environments
Enables advanced storage configuration
Add installation ID configuration
Support unique installation identifier for private deployments
Helps with telemetry and support for self-hosted instances
UI/UX Improvements
Improve in-app notifications
Add notification deletion and visual improvements
Makes it easier to manage and dismiss notifications
Overhaul Compare Dataset Versions view
Replace modal with full-page view, add checkbox selection and swap button
Makes dataset comparison more usable and feature-rich
Add move and remove actions to manifest modifications
Support removing or relocating elements in Kubernetes manifests
Provides more flexibility for manifest customization
Bug Fixes
Wait until inputs uploaded before creating workflow
Ensure file uploads complete before workflow run starts
Prevents workflow failures from missing input files
Fix copying pipeline with optional parameters
Properly handle optional parameters when copying pipelines
Ensures parameter controls appear correctly in copied pipelines
Fix copying pipeline not picking up environment variable groups
Include environment variable groups when copying pipelines
Maintains complete configuration in pipeline copies
September 17, 2025
Features & Enhancements
Optimize storage handlers for Google Cloud Storage
Improve performance of Google storage operations
Reduces latency when accessing data in GCS buckets
Add project and owner prefetch to metadata API
Optimize multi-metadata endpoint with eager loading
Speeds up metadata queries across executions
UI/UX Improvements
Enhance SME file upload fields
Add file preview, validation, and removal capabilities before upload
Improves user experience when preparing workflow inputs
Enhance client-side page titles
Update browser tab titles to reflect current page context
Makes it easier to identify tabs when multiple pages are open
Add allow-reuse configuration panel to pipeline task nodes
Show task execution reuse settings in pipeline editor
Provides visibility into caching behavior
Bug Fixes
Fix model version badge position in SME views
Correct positioning of model version indicators
Ensures proper layout in workflow configuration interface
Fix model input mapping in SME workflows
Properly map model version inputs when creating workflow runs
Prevents errors when workflows include model references
Handle long SME card descriptions
Prevent layout breaks with lengthy workflow descriptions
Ensures cards remain readable regardless of content length
Handle null model configuration in workflow runs
Gracefully handle cases where model input is not provided
Prevents errors when optional model inputs are empty
Use union query for task execution filtering
Optimize database queries when filtering by task with reused executions
Improves performance of execution list views
September 9, 2025
Bug Fixes
Disable task-reuse filter temporarily
Turn off task-reuse filtering in Executions table for performance reasons
Prevents slow queries while optimization work is in progress
September 9, 2025
Features & Enhancements
Use accessible project IDs optimization
Apply optimized query for retrieving accessible projects across more views
Improves page load performance for users with many projects
Bug Fixes
Avoid AlreadyChecking races when creating tasks
Prevent race condition errors during concurrent task creation
Ensures reliable task creation in high-concurrency scenarios
September 8, 2025
Features & Enhancements
Support default project and organization environments in notebook executions
Allow notebooks to inherit environment settings from project or organization defaults
Simplifies environment configuration for notebook sessions
Allow choosing notebook environments by slug
Support selecting notebook runtime environments using environment slugs
Provides more flexible environment selection options
Add workflow default inputs and parameters UI
Enable configuring default values for workflow inputs and parameters
Reduces repetitive data entry when launching workflows
Allow configuring exposed outputs for workflow definitions
Specify which pipeline outputs should be visible in workflow results
Simplifies result review for subject matter experts
Accept reuse-children property from task YAML
Support task execution reuse configuration in valohai.yaml
Enables declarative control over task result caching
Support custom Kubernetes manifests
Create Kubernetes API objects on-the-fly from custom manifest configurations
Allows advanced users to customize container orchestration
UI/UX Improvements
Dashboard: search projects by team
Add team-based filtering to project search
Makes it easier to find projects in multi-team organizations
Bug Fixes
Fix SME backlink using correct ID
Use proper identifier when navigating back from workflow run details
Ensures breadcrumb navigation works correctly
September 4, 2025
Features & Enhancements
Add SME user flagging
Enable marking users as subject matter experts with restricted interface
Provides simplified workflow-focused experience for non-technical users
Enhance SME workflow creation views
Improve project-level workflow creation and management interfaces
Makes workflow configuration more accessible to SME users
Add workflow management to project menu
Include workflow management section in project navigation
Improves discoverability of workflow features
Add time limit option for notebook executions
Allow setting execution time limits when creating notebook sessions
Helps control resource usage and costs for interactive workloads
Support limit in metadata filter
Add limit parameter to metadata query blocks for controlling result size
Enables more efficient queries when only top N results are needed
Add multiple metadata filter blocks
Support combining multiple metadata query blocks with union logic
Allows more complex filtering across different metadata criteria
Reuse task executions
Enable reusing previous task execution results instead of re-running
Improves pipeline efficiency by skipping redundant computations
Support library steps in pipelines
Allow referencing shared library steps using
library:...syntaxEnables code reuse across multiple pipelines
UI/UX Improvements
Add image preview beta improvements
Enhance image preview capabilities with better controls and zoom
Provides better visualization of execution output images
Add merge manifest modification
Support merging Kubernetes manifests for advanced deployment customization
Gives more control over container orchestration configuration
Add manifest modifications to deployment targets
Apply manifest modifications at deployment target level
Allows environment-specific Kubernetes configuration
Make basic TIFFs previewable
Add WASM-based decoder for previewing TIFF image files
Expands supported image formats in datum preview
Support custom resources on notebook executions
Allow requesting custom Kubernetes resources for notebook sessions
Enables specialized hardware like FPGAs for interactive workloads
Add Subtract differ mode to image comparison
Introduce subtraction blending mode for comparing images
Helps identify pixel-level differences between execution outputs
Bug Fixes
Prevent double-open of notebook tabs
Avoid opening multiple notebook tabs when clicking service buttons
Reduces confusion and browser tab clutter
Fix radio button and button hover styles
Restore proper styling for form controls after CSS updates
Ensures consistent appearance across the interface
Fix react-select stacking order
Correct z-index issues with dropdown menus
Prevents dropdowns from appearing behind other elements
Make "Stop" action confirm dangerous again
Re-enable confirmation dialog for stopping resources
Prevents accidental termination of running workloads
July 21, 2025
Features & Enhancements
Add Snowpark execution handler
Implement execution handler for running workloads in Snowflake Snowpark
Includes runtime configuration and UI for Snowpark environments
Add initial SME (Subject Matter Expert) models and API
Introduce workflow management system for subject matter experts
Provides simplified interface for non-technical users to run ML workflows
Support file upload and adoption for SME inputs
Allow SME users to upload files and adopt existing data as workflow inputs
Simplifies data preparation for workflow execution
Add workflow run model and creation API
Implement backend for creating and tracking workflow runs
Enables launching pipelines through simplified SME interface
Allow multipart upload for Oracle object storage
Support multipart upload protocol for Oracle Cloud Infrastructure storage
Improves performance when uploading large files to Oracle stores
UI/UX Improvements
Improve error messages when no global default store configured
Provide clearer guidance when storage configuration is missing
Helps users understand and resolve storage setup issues
Allow environment queue to be sortable by priority
Enable sorting execution queues by priority level
Makes it easier to manage and monitor high-priority workloads
Improve metadata query parsing for logical operators
Better highlight and parse AND/OR operators in metadata filter expressions
Reduces confusion when building complex metadata queries
Keep page selection on live output
Maintain current page position when viewing live execution logs
Prevents disruptive jumps to page 1 during log streaming
Bug Fixes
Fix execution admin panel
Add missing snowpark_job field to exclusion list in admin interface
Resolves errors when viewing executions in Django admin
June 18, 2025
Features & Enhancements
Support quoted strings in metadata queries
Allow using quoted strings in metadata filter expressions
Enables filtering by values that contain special characters or spaces
UI/UX Improvements
Move token policies behind organization token flag
Only show token policy management when organization has token feature enabled
Reduces UI clutter for organizations not using this feature
Bug Fixes
Fix pipeline crash when execution lacks cumulative metadata
Handle missing cumulative metadata gracefully in pipeline views
Prevents errors when viewing pipelines with incomplete execution data
June 18, 2025
Features & Enhancements
Improve token policy UI
Enhance usability of token policy creation and management interfaces
Makes it easier to configure security policies for API tokens
UI/UX Improvements
Format token expiry time in local timezone
Display token expiration dates in user's local timezone
Improves clarity when managing token lifecycles
Trim whitespace in metadata queries
Automatically trim leading and trailing whitespace from metadata filter queries
Prevents unexpected query failures from accidental spacing
Bug Fixes
Fix copying execution node environment variables
Properly copy environment variables when duplicating execution nodes
Ensures copied executions maintain their environment configuration
Return user-friendly error when store update fails
Provide clear error messages instead of generic failures during store updates
Helps users understand and resolve store configuration issues
Fix activity page handling inaccessible projects
Handle cases where activity feed includes projects the user can no longer access
Prevents errors when viewing activity history
Round environment price correctly in creation API
Fix rounding of per-hour pricing in environment creation endpoint
Ensures consistent price display across UI and API
June 11, 2025
Bug Fixes
Fix tolerations not read from valohai-yaml
Temporarily disable reading tolerations from YAML configuration
Prevents issues while double-step-name bug is resolved
June 11, 2025
Features & Enhancements
Add alert pings for executions
Support alert notifications from execution logs
Display alerts from log rate limiting directly in the log view
UI/UX Improvements
Allow organization admin to promote/demote team admin
Enable organization administrators to manage team admin privileges
Simplifies team role management within organizations
Bug Fixes
Revert valohai-yaml upgrade
Roll back valohai-yaml upgrade that caused issues with duplicate step names
Prevents configuration parsing errors in pipelines with step name conflicts
June 4, 2025
Features & Enhancements
Support project default environment for notebook executions
Allow notebook executions to use the project's default environment setting
Simplifies environment selection when creating notebook sessions
Add identity mapping configuration for executions
Enable configuration of execution identity mapping for enhanced security
Supports custom identity mapping in execution environments
Add tolerations to deployment endpoints
Support Kubernetes pod tolerations on deployment endpoint models
Allows deployment endpoints to be scheduled on specific nodes with taints
Initialize Snowpark environment type
Add support for Snowflake Snowpark execution environments
Enables running ML workloads directly in Snowflake compute
UI/UX Improvements
Improve duration tooltip formatting
Add detailed duration utility to show more precise time information
Provides better clarity for execution and pipeline durations
Allow team admins to edit team stores
Grant team administrators the ability to manage stores scoped to their team
Reduces dependency on organization admins for team-level store configuration
Show tolerations in cluster view
Display pod tolerations in deployment and execution cluster status views
Improves visibility into Kubernetes scheduling configuration
Bug Fixes
Fix team selection when creating or editing stores
Properly display teams selection dropdown in store creation and editing forms
Ensures team-scoped stores can be configured correctly
Last updated
Was this helpful?