> ## Documentation Index
> Fetch the complete documentation index at: https://arize-ax.mintlify.site/docs/llms.txt
> Use this file to discover all available pages before exploring further.

# On-Premise Releases

> Release notes for self-hosted Arize AX distribution images.

# Release 11.38.3 (2026-07-03)

## Updates

* Advance when no data error (#77170)
* Advance the last run on truncated window (#77108)
* Fix buffer truncated window advancement (#77189)
* Resolve CVE (#77627)
* Resolve CVE (#77503)
* Resolve CVE (#77638)

# Release 11.38.2 (2026-07-02)

## Updates

* Org admins cannot create a service key with org-admin permissions (#77235)

***

# Release 11.38.1 (2026-06-30)

## Updates

* Preserve newlines in highlighted code blocks (#77182)
* Adjust span delete interval timeout (#77175)

***

# Release 11.38.0 (2026-06-30)

## Updates

* Increase rate limits in EU (#75814)
* Add Alert to see when postgres connections are not being released (#75898)
* Add pagination param for listDatasetExamples endpoint (#72754)
* Datasets compare table hard to click into row (#75937)
* Make output on experiment run prettier for eval runs (#75933)
* Support Multi Region Deployment on AWS with AWS SDK v2 (#26849) OPEN
* Make the default span attributes display configurable (#71484)
* Support Data Fabric BYOB with on-prem (#70770)
* Update the username to match the SAML name attribute when existing user logs in (#73781)
* Hallucination Prompt Template Incorrect (#74519)
* Task logs UI does not surface skipped rows or errors, making failures difficult to debug (#72564)
* Can't filter on output column in dataset preview when setting up code evals (#74578)
* Add-user-to-space dropdown potentially capping available users, hiding users in large orgs (#74349)
* Deleting or annotating dataset records does not update `datasets.updated_at` (#75283)
* Querying for sessions is incorrect (#72280)
* Code eval fails with "Task was unable to handle template variable: documents" despite valid preview (#76120)
* Data is Truncated when Copying Examples to New Dataset (#74337)
* InferenceConfig Temperature Missing in AWS Bedrock Integration (#75775)
* TaskRuntimeError: Invalid column: attributes.input.value... errors occurring in online tasks when the attributes do exist (#75953)
* Input.value in eval variables mapping doesn't match the actual span attribute (#76483) OPEN
* Orgs not viewable after you hit 100 orgs (#76508)

***

# Release 11.37.1 (2026-06-24)

## Updates

* Update Netty (#75563)
* Correct filter dataset preview on experiment output column (#75896)

***

# Release 11.37.0 (2026-06-22)

## Updates

* cleanup azure MI docs (#73517)
* Support Multi Region Deployment on AWS with AWS SDK v2 (#26849) OPEN
* 502 error when deleting spans via Python SDK (#74705)
* Login Failed (#75490)
* Attentive: attributes.metadata.eval\_name overwritten to same value for all traces (#75615)

***

# Release 11.35.2 (2026-06-19)

## Updates

* Login failed after flood of space creations (#75490)

***

# Release 11.35.1 (2026-06-08)

## Updates

* Add internalendpoints bulk proxy (#74252)

***

# Release 11.35.0 (2026-06-08)

## Updates

* Update postgres username for adb control plane (#73552)
* Export per-service / per-endpoint REST API rate limits to docs (#73876)
* Bar chart bin logic creates double counting (#64812)
* Bad Transparent UI When Creating Drift Template Dashboard (#72332)
* Monitors page cannot scroll down (#74100)
* Generative-server not starting on IPv6-disabled clusters (#74013)
* Fix run\_experiment error in playground (#71886)
* Can't filter on tag when % Empty is very high (#72338)
* ADB UI config doesn't save empty decommissioned nodes list (#74045)
* Dashboard Bins are not being ordered correctly (#41501)

***

# Release 11.34.1 (2026-06-04)

## Updates

* Scale copilot memory (#73874)

***

# Release 11.34.0 (2026-06-02)

## Updates

* Ability to see date created and last usage on API Keys (#60313)
* Add value to set global max tasks for druidimporter (#71909)
* Include attributes.metadata.\* attributes in flight server span exports (#72532)
* Remove the ability to delete spaces by name (#72445)
* Add gemini-3.1-flash-lite to supported models (#72750)
* Expose source mapping config mutations (#72834)
* Allow Users to Add an Expiration Date to API Keys in the UI (#72522)
* Improve Metrics Tab Trace Graph View (#72666)
* Add kgateway example (#73046)
* Update postgres username for adb control plane (#73552)
* Unable to click update after adding a new eval to a task (#72467)
* Organization Admins can't get evals from private spaces in sdk (#72461)
* Prompts get - missing verbosity error (#72582)
* Eval Variable Mapping Column Selector (#72354)
* Revert fix tracing table height" (#72622)
* Dashboard tags flicker and conflict with UI element when applied (#72851)
* Post-login redirect drops monitor URL path and query params (#72331)
* Trace Workflow Doesn't Account for Time Window (#73187)
* Alyx Errors with Multi-Select (#73234)
* Evals Unexpectedly Added to Spans with "N/A" in label (#72610)

***

# Release 11.33.4 (2026-06-08)

## Updates

* Trace Slideover details not loading for late ingested spans (#74232)

***

# Release 11.33.3 (2026-05-29)

## Updates

* revert fix tracing table height (#72622)

***

# Release 11.33.2 (2026-05-29)

## Updates

* Dashboard Widget Account for Time Window (#73187)

***

# Release 11.33.1 (2026-05-26)

## Updates

* Expose source mapping config mutations (#72834)

***

# Release 11.33.0 (2026-05-21)

## Updates

* Ability to add other alerts to Pager Duty (#40027)
* Support Data Fabric BYOB with on-prem (#70770)
* Space Admins Cannot Add Members (#72321)
* UI user management unavailable despite disabled "sync permissions on each login" (#72444)
* Post-login redirect drops monitor URL path (#72331)
* Running Tasks + Tracing Projects Page Not Loading (#72249)
* UI restart required after on-prem upgrade (#72333)

***

# Release 11.32.3 (2026-05-20)

## Updates

* Resolve CVEs (#72040)

***

# Release 11.32.2 (2026-05-19)

## Updates

* Code Evals "Test Code on Example" does not complete (#71848)

***

# Release 11.32.1 (2026-05-18)

## Updates

* Add option to set clusterDomain (#72105)

***

# Release 11.32.0 (2026-05-16)

## Updates

* Enable realtime on all spaces on new installs (#70776)
* Update Druidloaderv2 Scaling Query (#71931)
* Enhanced Eval Visibility: Display Evaluation Status and Results in Log Tab (#59214)
* Online Tasks Continually get stuck in a pending state (#69565)
* Ingestion down and no pagerduty alert raised (#70820)
* Org Admins Unable to View Org or Space Members (#71707)
* Edit evaluator page issue when loading a project with typeahead dimensions (#71592)
* Evaluator edit page loads entire dataset (#69635)

***

# Release 11.31.1 (2026-05-14)

## Updates

* Online Tasks Continually in a pending state (#69565)
* Update Druidloaderv2 Scaling Query (#71931)

***

# Release 11.31.0 (2026-05-14)

## Updates

* Enable webhooks for all users (#70792)
* Inline Code Evals (#68912)
* Support base64 images larger than 16KB in traces table view (#70113)
* Update Terraform to have common tagging in AWS using default\_tags (#70715)
* Add reasoning/verbosity to LLM parameters for gpt-5 models (#60871)
* Comparison of the same time range on the perf dashboard yields different results (#44731)
* AX CLI v0.12.0 SSL regression for self-hosted/CloudFront deployments (#68477)
* request\_verify=false not respected on REST API code paths (#68454)
* Evaluator Variable Mappings page continuously calls SpansContentQuery (#70848)
* Dashboard "monitor summary" template not saving (#45173)
* export\_model\_to\_parquet fails when non-nullable columns contain nulls (#60320)
* Threshold field looks like it is editable when the monitor is not in edit mode (#55809)
* Contains Filter on Experiments Does Not Work (#70611)
* Task logs button doesn't work in the experiment comparison view (#71207)
* Settings Page API Keys Sort Does Not Work (#70810)
* Eval Tracing Toggle Not Saving (#71314)
* Non-admins can't access the account members page (#71535)
* Dashboards listing page cannot be scrolled (#71060)
* Bad formatting in the "pretty" input/output attributes display (#71487)

***

# Release 11.30.4 (2026-05-13)

## Updates

* Conditional apparmor annotation for inline custom code evals (#71594)

***

# Release 11.30.3 (2026-05-12)

## Updates

* Add internalendpoint req/resp log (#71337)
* `export_model_to_parquet` fails when non-nullable columns contain nulls (#60320)

***

# Release 11.30.2 (2026-05-07)

## Updates

* Evaluator Variable Mappings page continuously calls SpansContentQuery (#70848)

***

# Release 11.30.1 (2026-05-07)

## Updates

* Resolve CVE-2025-68121 (#70759)

***

# Release 11.30.0 (2026-05-07)

## Updates

* Support usage page with on-prem (#67038)
* Sign on-prem container images (#45971)
* Skip loading Stripe JS if onprem (#61020)
* Precheck postgres timezone/datestyle (#69216)
* Improve Alyx Error Messaging when Rate Limits are hit (#69963)
* Task run status stuck at running (#64054)
* Copy data easily in the compare experiments slideover (#61308)
* Add evaluate, langdetect to the custom code eval library (#70099)
* Alyx Generated Views (#69396)
* Evals on Datasets (#48761)
* Variable mapping issues (#69916)
* Eval Hover Modal linkage to Task Run (#65539)
* Expose mutation to set rate limits programmatically (#68295)
* Increment the default API version in Prompt Playground for Azure OpenAI (#59875)
* Additional span attributes aren't persisted in legacy code evals (#67380)
* Changes made to legacy custom code evals aren't persisted (#67381)
* Unable to scroll through all spaces in Custom Model Endpoint selection (#69307)
* Data uploaded to training environment without actuals is silently failing (#66881)
* Ranking Models - Rank showing up as a feature (#68642)
* Auto-add-to-dataset query filter parsing doesn't work with double quotes (#68938)
* Task Logs shows 0 for all counts on backfill eval tasks with multiple trace-level evaluators (#69567)
* Session filter incorrectly matches Boolean values for Custom Attribute despite 0 Count (#69641)
* Incorrect nodes and counts on the agent path (#69703)
* Experiment eval rows show score but no label/explanation when there's an error in the judge LLM response (#69632)
* Saved View Not Working Broken from Evaluator > Filter Sessions (#68234)
* Missing timestamp column in the performance dashboard table view (#63014)
* SAML role mapping mutation leaves organization field blank in UI despite successful backend update (#68621)
* New Org Not Propagated to Added Users Unless First Space is Created (#65131)
* View Experiment Traces returns unhelpful "An error occurred" (#70511)
* Log\_spans does not allow for custom attributes (#55015)

***

# Release 11.29.3 (2026-05-01)

## Updates

* Turn off the date validator (#69908)

***

# Release 11.29.2 (2026-04-30)

## Updates

* Auto-add-to-dataset query filter parsing with double quotes (#68938)

***

# Release 11.29.1 (2026-04-29)

## Updates

* Inline Code Evals (#68912)
* Importer parameter update (#67818)

***

# Release 11.29.0 (2026-04-29)

## Updates

* Support base64 images in playground that are not prefixed with data:image/ (#52538)
* Autosize generative server horizontally (#67732)
* Add Newer OA Models include O4 (#46958)
* Add developer checks to create api key mutations in app-server (#69715)
* Update service keys to do the deveoper checks in all places (#69852)
* Skip loading Stripe JS if onprem (#61020)
* Unable to successfully run evals on Base64 Images Without Prefix (#66884)
* List experiments endpoint returns 500: cannot unmarshal number into string for 'id' field (#68046)
* Fileimporter dialog keep reloading default schema (#69512)
* Annotation through SDK hits "Error logging arrow table to Arize AX" for permissions (#69540)
* Bedrock Throwing Error if using Custom Model Without Enabling Default Models (#69544)
* Updating dashboard widget with custom metric doesn't save (#69630)
* Dashboard text widget cut off after certain character limit (#67862)
* Cost Config Page Entries Duplicated (#69069)
* Alerts History API not showing past 30 days (#69854)

***

# Release 11.28.0 (2026-04-22)

## Updates

* Add SERVICE\_KEY\_READ and SERVICE\_KEY\_UPDATE permissions (#68442)
* ADB control plane and rate limits (#68782)
* Add a search to space selector (#67735)
* VertexAI API Error ECONNRESET (#69151)

***

# Release 11.27.1 (2026-04-21)

## Updates

* VertexAI API Error ECONNRESET (#69151)

***

# Release 11.27.0 (2026-04-21)

## Updates

* Expose API to pull aggregated tracing metrics (#68683)
* Improve migrate/restore from pg pod to cloud external postgres (#64806)
* Wrong order of annotation configs in the dataset row view (#64513)
* Sort the order of annotations on the projects page (#60651)
* Auto expand cells in dataset (#68695)
* Carry Over Column Selection in Datasets Page to Dataset Row Detail (#68736)
* Trace-level eval with span filter incorrectly shows "Running" state on traces missing the required span (#67941)
* Vertex Gemini 2.5-pro and 2.5-flash throwing 'exception posting request to model' error (#67287)
* "History" for Filter Bar in Experiments Page Does Not Work (#62202)
* Labelling queue - column selection resetting (#68567)
* Eval config shows false "Column not found" for nested JSON string paths (#68603)
* parentId is non-nullable on TaskRun GraphQL schema (#66616)
* Rate Limits cannot be Reset to No Limit (#68296)
* Vertex AI Claude Haiku 4.5 returns 404 due to incorrect model ID (#68734)
* Duplicate eval columns in dataset when spans already carry evals and a task re-evaluates the same column (#68909)

***

# Release 11.26.2 (2026-04-20)

## Updates

* Vertex Gemini 2.5-pro and 2.5-flash throwing 'exception posting request to model' error (#67287)

***

# Release 11.26.1 (2026-04-16)

## Updates

* Bedrock proxy\_with\_headers improvements (#68593)
* Custom Code Evals improvements for Playground (#65884)

***

# Release 11.26.0 (2026-04-15)

## Updates

* Fine Grained Access Control - Resource Level (#56430)
* Support realtime traces on existing spaces (#67916)
* Lift feature flag rbac (#68433)
* Respect user timezone setting on CSV export (#68315)
* Prompt Playground Azure OpenAI returns 400 for gpt-5.4-nano (#67221)
* Experiment compare slideover image render (#68239)
* Monitor Filter Saving (#68324)
* Incorrect Annotation Count on Example (#64578)
* Trace-level eval with span filter shows "Running" state on traces missing the required span (#67941)
* Login page 401 refresh (#67576)

***

# Release 11.25.1 (2026-04-12)

## Updates

* Respect user timezone setting on CSV export (#68315)
* Monitor Filter Not Saving (#68324)

***

# Release 11.25.0 (2026-04-11)

### Upgrade Notes

* ***values change***: gcpServiceAccountName and azureStorageAccountName are no longer base64 encoded

## Updates

* Enable count metric for non-numeric data in pivot tables (#62412)
* Enable Calico as a method for Network Policies in GKE in Terraform (#67559)
* Experiments page should render the rows with output != NULL filter pre-populated (#57195)
* Customize Rate Limit for Playground Runs (#67245)
* Link Back button on Dashboard to the Dashboards List Rather than Browser Page (#67821)
* Trace listing table shows blank input/output when detail view shows data (#65098)
* Alyx Issue When Asking Questions about Code Evals (#67096)
* Code eval ignores evaluator-level trace granularity — results merged at span level (#67575)
* Update does not update in Evaluator Version (#62656)
* Evaluator UI inconsistency Cannot remove LLM override once added (#65416)
* Dashboards - Unable to Duplicate Pivot Table Widget (#67847)
* Remove gcpServiceAccountName and azureStorageAccountName from secret (#65550)
* AX Profiles Create On Prem Throws Error When defining single host (#68040)
* Eval variables not mapping correctly (#68025)
* ArizeDB limit causes multi-span filters A and B yield different results than B & A (#65500)
* Playground run fails when eval is selected (#68126)

***

# Release 11.24.2 (2026-04-03)

## Updates

* Remove control plane alert rule (#67790)

***

# Release 11.24.1 (2026-04-02)

## Updates

* Trace listing table shows blank input/output when detail view shows data (#65098)

***

***

# Release 11.24.0 (2026-03-31)

## Updates

* Project-level RBAC for Tracing (#66110)
* Prevent phantom session eval scores (#67056)
* Sessions metrics bar (#67107)
* Persist time range in the tracing page URL (#67229)
* Cannot cancel task run — wrong RBAC permission checked (#66844)
* Session Evals Attached with "0" to Sessions the Eval Has Not Run On (#67050)
* Base64 Images are cutoff in experiments view (#65926)
* Playground Runs on More Rows than Dataset Has if Dataset has Duplicate IDs (#66593)
* Validation error during regex matching code eval setup due to undefined span attribute (#64925)

***

# Release 11.23.1 (2026-03-28)

## Updates

* Base64 Images are cutoff in experiments view (#65926)

***

# Release 11.23.0 (2026-03-27)

## Updates

* Gemini 3 Pro deprecation thru GCP Vertex (#66865)
* Expand Eval Avg Score in Experiments (#66952)
* Enable easy mapping between Arize AX Space IDs and Space UUIDs in Grafana (#66329)
* Azure managed identity for all services on on-prem (#48506)
* Improve migrate/restore from pg pod to cloud external postgres (#64806)
* "View Spans" from dashboard bar chart applies wrong series filters, returning 0 results (#66425)
* "View Spans" drill-down from dashboard chart resets time filter to 15m instead of inheriting selected range (#66423)
* Updating Eval Task target datasource from Project to Dataset doesn't update target in Eval Tasks tab (#65064)
* Task Rename Incorrectly Says "Task Run Started" (#65737)
* Evaluator rename not propagated to tasks and eval\_results (#65736)
* Dashboard-Wide Filters Do Not Apply to Distribution Widgets (#65652)
* Show monitor summary template on dashboard slideover (#66311)
* VertexAI API Error. Streaming is strongly recommended for operations that may take longer than 10 minutes (#66494)

***

# Release 11.22.3 (2026-03-26)

## Updates

* Gemini 3.1 models to Vertex AI provider (#66846)

***

# Release 11.22.2 (2026-03-20)

## Updates

* Internal server error on playground run (#65651)
* Bedrock integration sends empty externalid causing aws validation error (#63805)

***

***

# Release 11.22.1 (2026-03-17)

## Updates

* Copilot pod CrashLoop when using Azure Managed Identity (#65521)
* Alyx does not let user select a custom endpoint if model name is `gpt-4.1-custom` (#65785)

***

***

# Release 11.22.0 (2026-03-14)

## Updates

* Support Data Fabric BYOB on GCP/AWS (#54438)
* Support Gemini integration (#64478)
* Remove the index in the Experiments table (#65285)
* Support a custom pre-install for customer to collect secrets (#62902)
* Custom Code Evals not working in playground (#61815)
* Prompt Hub doesn't support out-of-the-box OpenAI tools (#64864)
* Postgres not available during the upgrade (#62747)
* Experiment eval labels are wrong colors (#65283)
* Custom time filter in Traces page when default language other than English (#57334)
* Evaluator multiselect deduplication and unclear naming (#65469)

***

# Release 11.21.1 (2026-03-10)

## Updates

* Setting retention on a datasource returns to global retention (#64585)

***

# Release 11.21.0 (2026-03-06)

## Updates

* Alyx 2.0 (#60212)
* Scope Better Graph Legends Readability (#41615)
* Bump arizedb-indexing indexing-volume (#63942)
* Add docs for in-cluster service reference (#62417)
* Update alermanager email subject format (#63440)
* Keep ML models turned on by default for new spaces (#64018)
* Enable filters in scatter plots (#62410)
* Honor client-sent cost (#62815)
* Add alermanager support for Teams, TeamsV2, Slack (#63339)
* Can't select the output column in prompt optimization task (#50493)
* Span Query Filters Incorrect When Selecting View Spans in Dashboard Distribution Widget (#63969)
* Wrong filter value applied for numbers like 35.1k when going from dashboards to traces (#53319)
* Dataset UI breaks on Chinese dataset (#58674)
* FILTER clause is incorrectly handling WHERE clause on list of strings (#63461)
* Incomplete filters applied going from a stacked bar chart to view spans (#62409)
* PSI Seems very low for the visualized distributions (#63862)
* Cardinality metric missing from dashboard widget metric dropdown for LLM models (#63630)
* Drift shows no data for training dataset when all data has the same timestamp (#64118)

***

# Release 11.20.1 (2026-02-28)

## Updates

* Add configurable custom IO parser (#64197)

***

# Release 11.20.0 (2026-02-23) (Maintenance)

## Updates

* Input/Output Table Rendering (#62223)
* Support custom model endpoint with bedrock API (#47740)
* Support for llm generation cost on playground, experiments (#50604)
* RBAC SAML UI (#63190)
* Tasks with multiple evaluators return NOT\_PARSABLE (#62760)
* Filtering on trace and session evals doesn't work (#62062)
* Cannot see nested attributes in eval variable preview (#62250)
* Can't use gpt-5-nano models for evals due to temperature settings (#60546)
* Online Evals - Gpt-5 max\_tokens error (#60031)
* Param Validation Error in Online Evals with Bedrock Sonnet 3.7, 4.5 (#61048)
* Incorrect routing applied from the time series chart hover (#62860)
* Can't scroll to see all session evals (#63173)
* DeleteSpace GraphQL call return "Success" regardless of success or failure (#61344)
* Building Dashboard Widgets is incorrect on Safari (#63148)
* Cost config table has same column names if applied to both prompt / completion columns (#63395)
* Missing AWS Bedrock Provider Params Validation Causes Task Failure (#62220)
* No traces with filter, but cost, tokens, latency still shows a value (#63015)
* Inputoutput table rendering (#62776)

***

# Release 11.19.2 (2026-02-19) (Maintenance)

## Updates

* Traces stats query causing incorrect data (#56226)
* Security updates (#63121, #63400, #63120, #63013)

***

# Release 11.19.1 (2026-02-12) (Maintenance)

## Updates

* Security update for grafana (#62875)
* Remove historicals OnRootMismatch (#62859)

***

# Release 11.19.0 (2026-02-12) (Maintenance)

## Updates

* Option to enable realtime traces for new spaces on GCP/Azure/AWS/Private Cloud (#43838)
* Support escaped curly braces in online eval templates (#61862)
* Generate java certificate from blobUserCertificate (#48207)
* ADB UI to replace existing console (#54676)
* Copilot support for LLM integrations (#62717)
* Add missing apiproxy discovery (#62415)
* Pred scores are incorrect when sending list of strings via java sdk (#61988)
* Annotations SDK does not support adding text annotations (#57343)
* dashboard\_time\_range\_key schema migration failure (#62735)
* Playground Erroring with Model Unreachable when Clicking "Function" (#62782)
* Tasks with multiple evaluators return NOT\_PARSABLE (#62760)

***

# Release 11.18.2 (2026-02-10) (Maintenance)

## Updates

* Users Still Able to Download Data via SDK with export disabled (#62580)

***

# Release 11.18.1 (2026-02-09) (Maintenance)

## Updates

* Parameter repoSubdir is not used in the chart/manifest (#61811)
* Parse json input and output attributes (#62386)

***

# Release 11.18.0 (2026-02-07) (Maintenance)

## Updates

* Option to enable realtime traces for new spaces on GCP/Azure/AWS (#43838)
* Support Azure Custom Endpoints with custom auth (#60430)
* Add traces to prober and datagenerator (#54981)
* Pivot table widget (#54533)
* Support not adding image pull secret when secret name is empty (#61810)
* Support Kubernetes 1.35 (#61880)
* Add Helm/ArgoCD hook to check operator reconcile status (#56917)
* Resolve CVE‑2025‑15467 (#62306)
* Session page unresponsive or crashing for very long sessions (#58881)
* Time selector doesn't work on sessions page (#59785)
* Copy+pasting text into playground doesn't update state (#61356)
* Responses Spans vs. Chat Completion Spans Differ in Image Attributes (#61194)
* Evals with more than 4 labels UI bug (#60890)
* Error when trying to log boolean labels via java sdk (#61983)
* Order in which we surface experiment results (#58918)
* Add missing rate limit controls (#62081)
* repoSubdir is not used in the chart/manifest (#61811) OPEN
* Clean up conditional PDB if defaultMinAvailable=0 (#59881)
* Granularity selection doesn't change drift timeseries chart (#60181)
* Configmap onprem-metadata creation issue (#62125)
* Annotation config names cut off at 30 characters (#61265)
* Dataset Evals Failing for Bedrock Models (#60862)

***

# Release 11.17.5 (2026-02-10) (Maintenance)

## Updates

* Users Still Able to Download Data via SDK with export disabled (#62580)

***

# Release 11.17.4 (2026-02-05) (Maintenance)

## Updates

* Add missing rate limit controls (#62081)
* Configmap onprem-metadata creation bug (#62125)

***

# Release 11.17.3 (2026-02-04) (Maintenance)

## Updates

* Fix add time to multi filter queries (#61812)

***

# Release 11.17.2 (2026-02-03) (Maintenance)

## Updates

* Evals with more than 4 labels UI bug (#60890)

***

# Release 11.17.1 (2026-01-30) (Maintenance)

## Updates

* Remove requirement for nodes/proxy from cluster role (#61640)

***

# Release 11.17.0 (2026-01-30) (Maintenance)

## Updates

* Evaluator Hub (#56888)
* Alertmanager - Webhook integration (#54427)
* Adding customSeccompProfile to allow null or any other profile configs (#61051)
* AWS Bedrock add baseUrl and custom models #60032
* Trace count wrong (#61245)
* Can't change the time range for an existing eval task (#61126)
* Session-scoped code evaluators fail to persist eval results - join column mismatch (#59847)
* Fix cgroup metrics for externalendpoints and onlinetasks (#61440)
* Unable to use \* in Custom Metrics (#59364)

***

# Release 11.16.0 (2026-01-27) (Maintenance)

## Updates

* Expose configurable timeout parameter for experiment task execution in run\_experiment (#59912)
* Auto-threshold training period greater than 30 days (#58097)
* Scatter plot widget (#54532)
* Expose User Creation Date in GQL (#60616)
* Add db fields for dashboard time persistence (#60784)
* Make time interval selector in Dashboards sticky (#46182)
* Persist global dashboard time range (#52339)
* Persist User-Selected Time Range and Granularity in Dashboard (#41614)
* Additional SAML UI search options (#59586)
* Add `conversation` Virtual Attribute for Session Evaluations (#57397)
* Support blocking data exports in flighserver, app-server and datacluster (#59281)
* Make postgres and gazette cleanup job retention period configurable (#59822)
* Option to run autosizer without node level metrics and cluster role (#53628)
* Rest API and GraphQL buttons are difficult to read in light UI theme (#60152)
* Can't apply a baseline filter in custom data quality monitors tracking cardinality (#58452)
* Drift Widget Changes Data when Title Changes (#57838)
* Filtering traces with apostrophe throws an error (#53098)
* Create space modal disappears when you try to add users to the space (#54143)

***

# Release 11.15.2 (2026-01-21) (Maintenance)

## Updates

* Enable arbitrary emails by default for onprem (#60589)

***

# Release 11.15.1 (2026-01-21) (Maintenance)

## Updates

* Update gazette image for newest ubuntu gpgv (#60463)

***

# Release 11.15.0 (2026-01-20) (Maintenance)

## Updates

* Expose SpaceSettingsPageServiceKeysTabContentDeleteAPIKeyMutation (#59555)
* Stacked bar chart widget (#54531)
* Add cost info to token tool tip (#59271)
* New Empty State for Evaluations Tab (#59268)
* New Trace Slideover Action Menu (#59270)
* Operator adds labels and annotations on all k8s resources, deprecates resourceLabel (#58823)
* Update alerting formatting text to be more readable (#59351)
* Add envoyproxy example for istio (#60057)
* Update java sdk to accept space ID in addition to space key (#54799)
* Preview table should extend the full length of page (#58920)
* ArizeExportClient hangs when no traffic exists within the queries period and the WHERE clause is used (#58001)
* Experiment Chart Metric Inaccurate/Inconsistent (#58673)
* Online tasks only running some evals not all (#56819)
* Annotations Rows Unclear in Slideover (#58045)
* Missing certificate in SAML metadata (#59995)
* Cannot filter on list attributes using contains operator (#54774)
* Normalize option doesn't work in distribution widgets (#60182)

***

# Release 11.14.2 (2026-01-12) (Maintenance)

## Updates

* Add llama 4 maverick (#59806)

***

# Release 11.14.1 (2026-01-10) (Maintenance)

## Updates

* Online tasks only running some evals not all (#56819)

***

# Release 11.14.0 (2026-01-08) (Maintenance)

## Updates

* Allow org admins to CRUD integrations (#56710)
* Restrict Email Input to Existing Account Members (#59080)
* Gazette journal replication config causing pod to not gracefully terminate (#58827)
* Dropdown Limit on Attribute Filtering in Dashboard Widgets (#59018)

***

# Release 11.13.0 (2026-01-02) (Maintenance)

## Updates

* Resize ML Table Columns (#58568)
* Use Experiments as datasource (#57911)
* Enable historical ephemeral clean-up in on-prem for flightserver data (#58948)
* Improve start/end node rendering with loopback on LangGraph Agent Graph (#56872)
* Eval Name blocks value (#58151)
* Unable To Toggle Space Roles In Public Space (#57492)

***

# Release 11.12.3 (2025-12-23) (Maintenance)

## Updates

* Handle the various error VertexAI can throw (#58855)

***

# Release 11.12.2 (2025-12-21) (Maintenance)

## Updates

* Update etcd image group (#58534)
* Fix Bedrock onprem support (#56397)
* Fix grant permissions on init (#58851)

***

# Release 11.12.1 (2025-12-19) (Maintenance)

## Updates

* Multi-span filters (#49652)
* Configurable threshold for historical PVC alert (#58453)
* Add Custom Integration option in UI (#57241)
* Disable Auto-Converting Monitor Times to Weeks from Days (#48700)
* Option to disable export traces (#58484)
* Add stop button functionality for small playground runs (#54232)
* GQL Mutation for Deleting Spaces (#58485)
* Search option when updating SAML configuration (#56907)
* Allow special characters in postgres password (#20728)
* Support direct updates to replace spark joiner (#55464)
* Experiments page should render the rows with output != NULL filter pre-populated (#57195)
* Unable to View/ Select Training Data After Ingestion (#49344)
* Monitors Running at Times Inconsistent with Downtime and Runtime (#47524)
* Wrong filter applied in dashboards "View spans" option to get spans with missing evaluation labels (#51770)
* Sonnet 4.5 in online evals fails because both top\_p and temperature are being passed in (#54176)
* Labelling queue column selection doesn't contain all columns from dataset (#55788)
* Cannot select all spans unless loaded in view (#52914)
* Unable to update eval unless you randomly click somewhere else before saving (#58582)
* Authorized spaces and organizations don't show up for the LLM provider integrations (#57094)
* Documentation Link Broken below Function Calling on Eval Edit Page (#58318)

***

# Release 11.11.1 (2025-12-16) (Maintenance)

### Updates

* Unable to View/ Select Training Data After Ingestion (#49344)

***

# Release 11.11.0 (2025-12-13) (Maintenance)

### Updates

* Customer requirements (#57185)
* Test config for xmatters (#56174)
* Add ability to specify topN via GraphQL mutation query to customize bar widget (#55204)
* Don't use 70 as user id for init jobs (#57779)
* Support 4.5 Haiku via Bedrock (#57438)
* Procedure to migrate from Arize AX postgres to cloud postgres (#56241)
* Validation errors coming before form populated on UI SAML Config (#57235)
* Need to set datastyle on external postgres dbs (#58209)
* Cannot apply Top N Binning Strategy due to Contradictory Requirements (#50388)
* Feature search sorted alphabetically , scroll doesn't include all entries (#53940)
* Filter by Contains String Does Not Work (#52215)
* Azure custom integrations with managed identity no longer work (#57761)
* Logging latent metadata for spans fails (#57086)
* Filters applied on baseline don't show up in the UI (#56068)
* Monitor triggers numbers incorrect - impacting summary dashboard (#56275)
* Sonnet 4.5 in online evals fails because both top\_p and temperature are being passed in (#54176)
* Can't create data quality monitors tracking cardinality - new values (#57915)
* Fix error message to only return unique span ids (#55619)
* Downloading spans from the UI only downloads the last batch if download size is >1000 (#56906)

***

# Release 11.10.7 (2025-12-11) (Maintenance)

### Updates

* Cannot see traces from monitor (#58131)
* Members are Unable to Create Service Key (#57643)
* Online task support eval level llm config

***

# Release 11.10.6 (2025-12-08) (Maintenance)

### Updates

* Autosizer scales below replicaMin annotation defined (#56894)
* Update go-migrate for on-prem (#58092)
* Update fileimporter alert to avoid false alerts (#58030)

***

# Release 11.10.5 (2025-12-05) (Maintenance)

### Updates

* Increase multi class class limit to 500 (#57731)
* Add update saml idp to the api schema (#57758)

***

# Release 11.10.4 (2025-12-05) (Maintenance)

### Updates

* Fix special characters in pg password (#57966)

***

# Release 11.10.3 (2025-12-05) (Maintenance)

### Updates

* Move pg-jobs to toolbox image (#57925)
* Add quotes around postgres reserved keywords (#57905)

***

# Release 11.10.2 (2025-12-03) (Maintenance)

### Updates

* Azure custom integrations with managed identity (#57761)
* Change min cpu/mem for operator (#57698)

***

# Release 11.10.1 (2025-12-02) (Maintenance)

### Upgrade Notes

* ***Breaking Change***: If using custom endpoints, review with Arize AI.

### Updates

* Option to specify image suffix (#57473)
* Change min cpu and mem requests (#57359)

***

# Release 11.10.0 (2025-12-01) (Maintenance)

### Upgrade Notes

* ***Breaking Change***: If using custom endpoints, review with Arize AI.

### Updates

* Integrate apiproxy with on-prem (#56939)
* Structured outputs in Prompt Hub (#52460)
* add option to pass in secret name instead of cert/key for internalEndpointsAppTlsCert/internalEndpointsAppTlsKey (#55748)
* Add timeout option to arize.sh (#56916)
* APIStatusError: Support for non-CONNECT requests not implemented - HTTP/POST (#53907)
* Wrong LLM integration of that same provider gets activated in Playground during model selection (#57143)
* Traces disappearing in Session View when Updating Annotations (#57216)
* Experiment Metrics Positive Class Shows Empty or Null (#57184)
* Broken Link from Create Eval Template 'See Documentation' (#56079)
* Dataset CSV Download is Missing Annotation Values (#55014)
* Custom code evaluator TypeError: write\_result\_metrics\_to\_fileargument count mismatch (#57263)
* Customer Seeing Error When Using Custom Metrics (#57391)
* Navigating back to perf tracing page after clicking on feature details results in no data error (#57290)

***

# Release 11.9.1 (2025-12-03) (Maintenance)

### Updates

* Reduce page max records to address UI slowness (#57778)

***