v2.38.2-odh

Notable Changes

• New features:
  • Enhancements to Model Catalog Search & Filtering with Model Validation Data
  • Enhancements to Model Details with Model Validation Data
  • AI Available Assets - LLS Deployed Models, MCP Servers (via Config-Map)
  • Feature Store and workbenches integration
  • Hardware Profiles is now GA
  • Model mesh is not supported anymore
• Ongoing features
  • New flow for Model deployment
  • Dashboard BYO OIDC Impediments
• Miscellaneous bugs and improvement

What's Changed

• Added a warning for Legacy NVIDIA API keys when enabling NIM by @TheiaSurette in #4824
• Fix row expansion issue after reordering by @nananosirova in #4940
• adjust sort order of model registry navigation items by @christianvogt in #4941
• chore: Add pgvector resource tile by @dgutride in #4920
• add handler for ApiPathPrefix by @pnaik1 in #4943
• Add Cypress E2E GitHub Actions workflow by @antowaddle in #4917
• fix: cypress workflow trigger only on /run-e2e-tests-gh comments by @antowaddle in #4947
• Assign node selectors and tolerations to InferenceService when an accelerator/legacy hardware profile is selected by @nananosirova in #4833
• Sync from kubeflow/model-registry bbb91eb by @mturley in #4948
• fix: Projects being fetched on every route change by @claudialphonse78 in #4899
• minor title textual change for data source and node issue resolution fix by @claudialphonse78 in #4958
• Fix loading issue in deploy from MC by @ppadti in #4957
• Frontend Streaming Changes by @ikeola13 in #4945
• Cypress e2e Pipeline Final Tweaks & Additional Cypress e2e Maintenance by @antowaddle in #4950
• init llmd model serving package by @emilys314 in #4949
• feat: left navigation reorganization and page updates by @caponetto in #4867
• Show llmd deployments alongside other platforms by @emilys314 in #4963
• Add Num Replica Section to Deployment Wizard by @ashley-o0o in #4903
• Sync from kubeflow/model-registry 0f52126 by @mturley in #4966
• lint frontend and add node lint plugin by @christianvogt in #4884
• Add LSD configuration endpoint and simple modal by @DaoDaoNoCode in #4933
• Add override for rhoai for modular architeture by @lucferbux in #4969
• Follow up to add token auth by @ashley-o0o in #4888
• add scripts for running dev e2e by @christianvogt in #4936
• fix:added scroll into view for lineage and style fixes for the toolbar by @claudialphonse78 in #4975
• Fix re-render loop by @nananosirova in #4964
• fixed axios cve by @rsun19 in #4980
• add default vscode settings to configure jest runner with typescript support by @christianvogt in #4981
• set default retries for mock tests to 2 by @christianvogt in #4985
• update mf shared config by @christianvogt in #4942
• fix:added review changes for data sources page by @claudialphonse78 in #4992
• feat: override MR/MC upstream page headers according to the mocks by @caponetto in #4974
• Fix optional chaining for access mode settings in StorageClassEditModal and utils by @Gkrumbach07 in #4996
• fix HMR for start:dev:ext by @christianvogt in #4997
• update port to 80 by @pnaik1 in #4971
• Sync from kubeflow/model-registry bcc518b by @mturley in #4986
• Disable Model Mesh v3.0 by @andrewballantyne in #4973
• Add networking to allow connection in rhoai by @lucferbux in #4995
• Add models list view in assets page by @divyanshiGupta in #4906
• Parametrize oauth-proxy image by @lucferbux in #4939
• RHOAIENG-27684: The workbench image column shows Deleted label after RHOAI upgrade by @srtanish1992 in #4988
• Remove legacy hardware profiles by @jrenee42 in #4879
• add unit test for hooks and api by @pnaik1 in #4993
• Added temperature and top_p sliders, added unit tests by @ikeola13 in #4990
• feat: update page headers to match their nav entry by @caponetto in #5010
• fix start:dev:ext with psi clusters by @christianvogt in #5020
• RHOAIENG-34758: Update MCP Default NS to Dashboard NS by @Lucifergene in #4951
• RHOAIENG-33403,RHOAIENG-32290,RHOAIENG-33405: MCP Tab in AI Assets + MCP Chat Panel + Tool Response (UI) by @Lucifergene in #4883
• Add Args and Parameters Section to Deployment Wizard by @ashley-o0o in #4979
• Remove 'disableFineTuning' field from odhdashboardconfigs CRD definition by @Gkrumbach07 in #4976
• update: upgrade Axios to version 1.12.0 across all packages by @bobbravo2 in #5024
• Change overlay name in rhoai by @lucferbux in #5029
• [feat] Add status of model for AAA page by @varshaprasad96 in #4953
• Add description field to deployment wizard by @katieperry4 in #5002
• chore:version upgrade for 2.25 by @claudialphonse78 in #5035
• Support more llmd viewing data by @emilys314 in #5005
• Starting a standalone workbench with a hardware profile yields an error by @nananosirova in #5001
• RHOAIENG-33400: Restructure BFF APIs and file organization by @Schimuneck in #5040
• Remove group errors and allow freeform group names by @andrewballantyne in #5034
• Update notebook creation to support new Connections API by @Griffin-Sullivan in #4771
• Add Available AI Assets Checkbox and Use Case fields by @katieperry4 in #4965
• Swap Admin Group Reading/Detection for SSAR of Auth by @andrewballantyne in #5028
• RHOAIENG-34215: Remove the pipelines option from hardware profile visibility settings by @srtanish1992 in #4968
• dw permission ux fix by @andrewballantyne in #5047
• [fix] Add check to ensure we don't install LSD twice by @varshaprasad96 in #5039
• Add Serving Runtime Select to wizard by @Griffin-Sullivan in #4982
• fix: add agents api as it includes responses by @Ygnas in #5051
• Moving saveAsAiAsset up a level by @katieperry4 in #5048
• feat(gen-ai): Add MaaS models and tokens APIs to BFF by @rohitkrai03 in #5042
• GEN AI - Add playground configuration modal by @DaoDaoNoCode in #5013
• Moved npm-run-all to dev-dependencies by @ikeola13 in #5056
• [feat] Add SA token while fetching models by @varshaprasad96 in #5033
• fix: broken model urls and failure to get inference service by @Bobbins228 in #5043
• fix package-lock check and workflow cache by @christianvogt in #5011
• updated wedbpack to fix deployment issue by @ikeola13 in #5060
• Add tests, edit, and deploy functionality for wizard model location by @katieperry4 in #4891
• RHOAIENG-32557-32558-32559: Replace DSP or Data Science Pipelines with Pipelines (UI only) by @srtanish1992 in #5044
• Add model filter by @Bobbins228 in #5059
• Add support for adding models to playground by @divyanshiGupta in #5057
• Remove Serverless by @andrewballantyne in #5021
• feat: refactor navigation routes by @caponetto in #4983
• Add Delete Action for LLMd by @ashley-o0o in #5052
• feat: Add file and vectorstore delete/list operations with query para… by @Schimuneck in #5046
• feat: Add previous response ID support for conversation continuity by @akram in #5063
• [feat] Add support for llminferenceservice by @varshaprasad96 in #5058
• Remove tests quarantine2nd oct2025 by @antowaddle in #5069
• Remove delete button for the OOTB hardware profile by @nananosirova in #5065
• feat: Add temporary username-based vectorstore isolation and file enr… by @Schimuneck in #5073
• GEN AI - Add delete playground and update configuration flow by @DaoDaoNoCode in #5066
• Add model ID while returning model list in AAA page by @varshaprasad96 in #5077
• Feature/multidoc upload by @ikeola13 in #5068
• Add hardware profile binding state labels for notebook and model serving tables by @nananosirova in #4897
• fixed rag toggle bug by @ikeola13 in #5081
• Create Github Action to check manifest generation by @lucferbux in #5030
• Create llmd deployments by @emilys314 in #5062
• Deprecate notebookSizes, modelServerSizes, notebookTolerationSettings, AP and HWP feature flags and add validations by @nananosirova in #5072
• Update HardwareProfile API version to v1 across mock data and tests by @Gkrumbach07 in #5075
• Fix for multiple a11y violations by @agagancarczyk in #5086
• Cypress 3.0 Test Maintenance by @antowaddle in #5085
• Fix llama model and aaa model matching issues by @DaoDaoNoCode in #5082
• Fix gen-ai routes by @divyanshiGupta in #5088
• Sync from kubeflow/model-registry de010c4 (fix /ai-hub/* paths) by @mturley in #5084
• Cleanup assemble inference service and add loading states by @Griffin-Sullivan in #5079
• Show model id in model info popover by @divyanshiGupta in #5087
• Removal of top_p field from model details by @agagancarczyk in #5097
• fix: vllm model port being wrong by @Ygnas in #5096
• [feat] Add SSAR to verify if user has access to LSD service by @varshaprasad96 in #5098
• Added more ui enhancements by @agagancarczyk in #5099
• Add creating an LLMD deployment test by @emilys314 in #5091
• Update nav icons and page header icons by @caponetto in #5101
• Remove platform selector from serving runtime template create page by @Griffin-Sullivan in #5092
• feat: add BFF build support for OpenShift by @akram in #5107
• Added mcp tools for the code exporter by @Ygnas in #5071
• Delete cypress logs by @FedeAlonso in #5110
• fix: hide navigation sections with no accessible items by @caponetto in #5111
• Update cypress mocks for wizard by @sridarna in #5094
• Gen ai UI/fixes by @ikeola13 in #5108
• [feat] Add wiring for llm-d install by @varshaprasad96 in #5109
• fix(backend): increase file upload limit by @rohitkrai03 in #5133
• Fix hardware profile updating in the expandable section of deployment table by @nananosirova in #5127
• Add maas models list in AA endpoints page by @divyanshiGupta in #5112
• Deprecate hwp smallest changes possible by @jrenee42 in #5053
• fix: add missing header icon when there is no project available by @caponetto in #5136
• RHOAIENG-35753: Display Streaming Errors in the Playground by @Lucifergene in #5138
• [feat] Implement static token management for inferenceservice by @varshaprasad96 in #5144
• add llama stack operator as required component by @pnaik1 in #5128
• Model Location 1C Connection Creation by @katieperry4 in #5064
• feat(maas): add MaaS feature flag and ai assets tab extension by @rohitkrai03 in #5148
• Deploy gen ai architecture with odh operator by @pnaik1 in #4877
• [feat] Implement static token management for LLMInferenceService v1alpha1 by @ChristianZaccaria in #5147
• fix: inference headless service port resolution by @Ygnas in #5153
• Add missing kebab actions to MR and other UX fixes by @manaswinidas in #5050
• [feat]: add SubjectAccessReview validation to all namespace-requiring endpoint by @varshaprasad96 in #5152
• fix: increase file upload timeout for large PDFs by @Schimuneck in #5157
• fix: RHOAIENG-35921: Update Vector Stores List to use updated LlamaStack API integration by @akram in #5155
• Fix Latest 3.0 Cypress E2E test Failures by @sridarna in #5150
• Update deployment secret creation name source by @katieperry4 in #5160
• Gen ai/chat UI fixes by @ikeola13 in #5161
• Adjust how token auth works with llmd deployments by @ashley-o0o in #5093
• Remove pagination, disable row dragging during filtering, remove column sorting in hardware profiles table by @nananosirova in #5154
• Add model location to llmd deployment by @katieperry4 in #5121
• Fix conditional hook invocation in useHardwareProfileBindingState by @nananosirova in #5173
• Add kserve as dependency for llmd by @emilys314 in #5166
• Avoid importing util directly from MR in serving by @mturley in #5167
• first round: accelerator profile dead code removal by @jrenee42 in #5140
• Increased temperature slider max value by @ikeola13 in #5175
• [feat] Support MaaS during LSD install by @varshaprasad96 in #5164
• Set model type to mandatory by @ashley-o0o in #5165
• Add error alert on failed submit by @emilys314 in #5169
• Remove dashboard hardware profiles CRD by @nananosirova in #5100
• Add existing resource and hardware profile resource merging logic in useHardwareProfileConfig by @nananosirova in #5172
• Modify oci secret details file field in test variables file by @sridarna in #5143
• Show AI Assets when deleting/stopping a model by @ashley-o0o in #5159
• Sync from kubeflow/model-registry 06a2015 by @mturley in #5176
• feat: add flag for custom distro names for lsd install by @Bobbins228 in #5163
• fix: Update LSD install swagger example to show correct model format by @akram in #5185
• Update maas models empty state by @divyanshiGupta in #5187
• [RHOAIENG-32843] Feature Store Workbench Integration by @claudialphonse78 in #5149
• [fix] Remove safety provider from configmap and verify if token exists by @varshaprasad96 in #5171
• chore: Fix swagger for MaaS endpoints by @varshaprasad96 in #5189
• Feature/model provider custom headers by @Schimuneck in #5141
• Hide wizard summary page by @ashley-o0o in #5177
• Remove remaining notebook only tolerations logic by @nananosirova in #5156
• Microcopy changes for AI asset endpoints tab by @agagancarczyk in #5195
• Add MaaS checkbox with wizard form field extension modifiers by @emilys314 in #5116
• add support for multiple filters by @pnaik1 in #5196
• Modify LSD install api - ui by @ikeola13 in #5188
• Implement MaaS Service autodiscovery in Gen-AI BFF by @ChristianZaccaria in #5178
• Follow up fixes for GEN AI Playground UI by @DaoDaoNoCode in #5184
• 32494/wizard microcopy updates by @ashley-o0o in #5207
• LLMD Editing in the Wizard by @katieperry4 in #5124
• Add event tracking for GEN AI playground by @DaoDaoNoCode in #5142
• Update kserve deployment by @emilys314 in #5181
• Maas model fetch improvement by @DaoDaoNoCode in #5214
• Sync from kubeflow/model-registry e67201d by @mturley in #5199
• feat: Propagate Authorization Bearer token to OpenAI client by @akram in #5216
• refactor: Convert LlamaStackConfig from string template to struct by @akram in #5215
• chore: filter nomic embed model gen ai playground by @Bobbins228 in #5234
• update splitLlamaModels by @pnaik1 in #5230
• Added try in playground button by @ikeola13 in #5235
• Notebooks annotations by @jstourac in #5162
• Fix auth token management in the model deployment wizard by @emilys314 in #5217
• Updated microcopy for playground by @agagancarczyk in #5237
• Fix location clearing by @katieperry4 in #5239
• Fix the pre-defined args/variable tooltips by @ashley-o0o in #5238
• Use kubeRBACProxy instead of oauthProxy in the spec when creating/editing modelregistries by @mturley in #5241
• migrate usage of dsc installedComponents to managementState by @christianvogt in #5183
• Add manifest for kube RBAC enablement by @lucferbux in #5076
• Quarantine-model-serving-tests by @sridarna in #5240
• Add MaaS model to check the llama model enabled status by @DaoDaoNoCode in #5244
• support gateway and kube-auth-proxy with frontend start:dev:ext by @christianvogt in #5243
• Add llamastackoperator to mapping by @DaoDaoNoCode in #5247
• Enable deployment wizard and microcopy stuff by @emilys314 in #5245
• chore: improve gen-ai bff http client logging for enhanced observability by @Bobbins228 in #5170
• fix: Handle provider-prefixed model IDs in GetModelProviderInfo by @Schimuneck in #5233
• Chore: Add RAG to tool groups by @varshaprasad96 in #5255
• fix(cypress): Handle OAuth login flows in visitWithLogin command by @antowaddle in #5252
• modified max chunk length and chunk overlap by @ikeola13 in #5258
• Cypresse2e3.0 maintenance21 oct by @antowaddle in #5261
• installed components to management state migration by @ugiordan in #5251
• Add GitHub action to sync odh-release with main by @manaswinidas in #5265
• Revert 85a6f35 by @christianvogt in #5274
• override form-data version by @christianvogt in #5270
• Update Tekton files to version v2.38.2-odh by @odh-devops-app[bot] in #5277
• Upversion dashboard by @manaswinidas in #5278

New Contributors

• @Bobbins228 made their first contribution in #5043
• @jstourac made their first contribution in #5162

Full Changelog: v2.38.1-odh...v2.38.2-odh