Skip to main content
Version: v1.2.x

Release Notes

August 7, 2026 — PaletteAI 1.2.1​

PaletteAI 1.2.1 — Patch release on 1.2.0: hotfix.1 image train that isolates Project spoke definitions from sibling Projects, keeps namespace stubs in sync as Projects change, and hardens federation against missing references. Workload scheduling is more reliable with co-location contention fixes, day-2 worker pool handling, and OCM Placement tolerations, and Workload status updates faster through watch-based feedback. Model and tenancy settings are picked up promptly by AIWorkload and ComputePool informers, Project settings are gated on the projects:update permission, and a broad set of console issues are resolved across scopes.

July 29, 2026 — PaletteAI 1.2.0​

PaletteAI 1.2.0 — First stable release in the 1.2 line: system-wide configuration and federation across Tenants and Projects, a new cross-scope sharing model for Apps, Models, Compute Pools, Compute Configs, Integrations, and Scaling Policies, an automatic Curator spoke agent for hardware inventory, customizable RBAC roles, non-default NIC and static IP configuration, control plane-only clusters, STIG security and compliance hardening (login notice-and-consent, session inactivity timeout, and more), GPU variants with Day-1/Day-2 MIG partitioning, and support bundle collection via CLI.

July 15, 2026 — PaletteAI 1.1.8​

PaletteAI 1.1.8 — Patch release on 1.1.7: hotfix.7 image train with overwrite: false on the legacy prometheus-agent add-ons, per-GPU gpu-memory semantics, reliable hub variable resolution, and hardened brush namespace RBAC and tenant federation deduplication.

July 10, 2026 — PaletteAI 1.1.7​

PaletteAI 1.1.7 — Patch release on 1.1.6: hotfix.6 image train that enables Tenant-scoped App and Model Deployments by default, grants default Dex users permission to create Tenants, hardens Compute Pool and control-plane scheduling, improves single-node cluster handling, adds paletteai support-bundle and --extra-pack, and improves Profile Bundle sharing and mirror sync.

June 30, 2026 — PaletteAI 1.1.6​

PaletteAI 1.1.6 — Patch release on 1.1.5: hotfix.5 image train with Prometheus remote-write auth, paletteai mirror --extra-repo and --extra-artifact flags, Tenant workload namespace federation, Canvas sharing and Tenant settings fixes, ProfileBundle sync improvements, read-only root filesystem for all containers, Zot TLS at Traefik LB, and automatic drift remediation for worker pool metadata.

June 18, 2026 — PaletteAI 1.1.5​

PaletteAI 1.1.5 — Patch release on 1.1.4: hotfix.4 image train, adding Compute Pool node pool taints, Compute Pool scale-up and overview improvements, Definition and Profile Bundle correctness fixes, Workload lifecycle and Settings validation hardening, paletteai CLI version and mirror --extra-chart additions, and numerous UI fixes.

June 6, 2026 — PaletteAI 1.1.4​

PaletteAI 1.1.4 — Patch release on 1.1.3: hotfix.3 image train, with Compute Pool lifecycle robustness, static overlay IP networking, definition and OCI sync correctness fixes, Tenant-scoped Profile Bundle import and Models list, and numerous Workload Profile and UI fixes.

May 29, 2026 — PaletteAI 1.1.3​

PaletteAI 1.1.3 — Patch release on 1.1.2: hotfix.2 image train, with Tenant permission and routing improvements, Tenant-scoped App Model deployment, Compute Pool health and naming fixes, Workload Profile and Profile Bundle handling fixes, and Project model settings scoping.

May 23, 2026 — PaletteAI 1.1.2​

PaletteAI 1.1.2 — Patch release on 1.1.1: hotfix.1 image train, WorkloadDeployment health when Environment is unready, AIWorkload and ComputePool status condition fixes, ProfileBundle finalizer and WorkloadProfile verification fixes, Helm cleanup reliability, and UI table layout fix.

May 21, 2026 — PaletteAI 1.1.1​

PaletteAI 1.1.1 - Patch release on 1.1.0: hotfix image train, Compute Pool health status improvements, spoke tenant definition federation fix, canvas and gamut updates, and new hue.tenantName for spoke OCI sync.

May 14, 2026 — PaletteAI 1.1.0​

PaletteAI 1.1.0 — Stable GA for the 1.1 line since 1.0.7: Traefik as default ingress, tenant-wide settings and GPU governance, Profile Bundle and deployment workflow improvements, paletteai mirror for airgapped installs, and all intervening 1.1.0-rc component releases. Refer to the full page for breaking changes, upgrade steps, Helm values.yaml diff, and per-component release notes.