What's New in Altair AI Hub 2026.0.0?

Released: September 1, 2025

The following describes the enhancements and bug fixes in AI Hub 2026.0.0:

AI Hub Server

Improve visualization of Endpoint test results inside the front-end which includes clear indicators for recently added arbitrary file input and output results feature

❗ breaking Return 204 for successful process invocations which yielded no IOObject as output for Scoring Agent service responses in contrast to returning an empty table and remove enforcing Accept header check for custom defined content-type annotated IOObjects
Add handler for NoResourceFoundException thrown by Spring for missing actuator endpoints
Add thread name prefix and group name configuration for task schedulers
Agents are only serving when their status is UP
Allow filtering for state and group in Projects Endpoints tab
Avoid tool tip and dialog overlap in operations page
Enhance logging for old deployment zip cleanup
Generate a database ID for archived logs when they have not been retrieved yet
Improve media type compatibility check in ExecutionService
Make queue filters case insensitive

❗ breaking Increase required feature version to 26.0 from 22
- This requires a new 2026 issued Altair license for on-premise Altair License Managers (ALM)
- This doesn’t require an upgraded on-premise ALM itself, but at least ALM version 15.5 to be in line with the internally used licensing framework library

Add handler for NoResourceFoundException thrown by Spring for missing actuator endpoints

Added a feature for user defined custom locations in the reverse proxy
Added a sample user defined location template to expose ActiveMQ WebUI (disabled by default)
Added timeout settings for Grafana, Grafana proxy and WebAPI components
Replaced busybox container reference on the dockerhub image to rapidminer-busybox for the ActiveMQ init container service
Refactored Custom Certificate Authority related documentation to meet the strict verification standards introduced by urllib3 v 2.4.0
License Proxy hostname can now be customized in the configuration to support identification in case of multiple deployments.

Fixed LicenseProxy health endpoint and turned on metrics exposure by default
Replaced gid check logic in RapidMiner proxy component to display a warning in the logs instead of exiting with error code

Upgraded Keycloak to 26.3.3
Upgraded Grafana to 12.1.1
Upgraded Jupyterhub to 5.3.0-15
Upgraded Git-LFS 3.7.0
Upgraded ActiveMQ to 2.42.0
Upgraded NVidia Cuda version in our GPU images to 13.0
Upgraded Python 3.13.7
Upgraded MonetDB Mar2025-SP1
Upgraded postgres from 14 to 17 for keycloakdb, aihubdb, jupyterhubdb (Requires migration)
Postgres migration from 14 to 17 can be delayed, we keep patthcing and releasing the 14 versions as well.

This pages below explain how to upgrade AI Hub PostgreSQL services from version 14 to version 17 using pg_dump and pg_restore.

According to your deployment, visit the Kubernetes or the Docker guide for details.