On-Premise Releases
Release 10.21.0 (2025-09-02)
Bug Fixes
Update calls to GQL Client to match execute method signature or pin version < 4.0.0 (#51466)
Release 10.20.2 (2025-08-28)
Bug Fixes
Operator UI does load (#52029)
Release 10.20.1 (2025-08-28)
Bug Fixes
Custom code evals fail with no such file or directory (#51978)
Release 10.20.0 (2025-08-27)
Features
Highlights 08.2025
Expose public graphql agent network query (#51675)
Add option to configure postgres username (#51775)
Bug Fixes
Missing eval scores on experiment (#50434)
Incompatible input format for Anthropic models using Bedrock integration (#50203)
No logs/errors visible on a failed prompt optimization task (#50583)
Release 10.19.0 (2025-08-21)
Features
Add pydantic to allowed packages for custom code evals (#51516)
Bug Fixes
Get model URL graphql query returns SaaS link for VPC deployments (#50789)
Annotation updated_at timestamp units are inconsistent on datasets depending on creation method (#51443)
'Preview' in Labeling Queue Does not Render Markdown (#50732)
Custom metrics not rendering (#51245)
Release 10.18.0 (2025-08-19)
Features
Log SAML logins to audit log table (#50782)
Unable to See Parent ID In Span (#48435)
Allow external Postgres port to be configured (#51216)
Release 10.17.0
Features
Smarter Version Selection (#47678)
OTLP Endpoint Rejects Multi-Resource Span Batches (#50026)
Bug Fixes
Initial configuration fails on AWS Aurora version 7.5 (#51162)
Selecting the same model twice in prompt playground, errors out (#50344)
Prediction group ID is required to be numeric for custom NDCG metric (#49720)
Release 10.16.0
Features
Configure request rate in onlinetasks (#50727)
Append CSV to an existing dataset (#44166)
Bring Copy IDs throughout core product entities (#50611)
Move Agent Graph to a Separate tab on the tracing slideover (#50639)
Bug Fixes
"Learn More About Roles" Link Leads to "Page Not Found" on Invite User Workflow (#50609)
Dataset filter history doesn't save (#50223)
Filtering on experiments is may not work (#50265)
Release 10.15.2
Features
Minor cosmetic changes (#46436)
Release 10.15.1
Features
Configure request rate in onlinetasks (#50727)
Release 10.15.0
Features
Provide guidelines for Karpenter (#47235)
Let users specify secrets to include in secret-manager-policy (#50248)
Make annotation configID visible in the UI (#50212)
Improve Azure OpenAI Custom Endpoint support validation (#49701)
Extend baseline lookback for drift widgets to > 1 year (#50394)
Allow Custom Model Integration through HTTP (#49733)
Support Images on Datasets & Labelling Queues (#49582)
Improve AWS Terraform - AMI, launch template, EBS encryption (#50050)
ArizeDB support for Self-Hosted (#48827)
Bug Fixes
AWS Bedrock request fails from Azure cluster (#49488)
When trying to update dataset with empty dataframe the datasets client hangs (#50259)
Sessions page refetches until end of data (#50350)
Duplicate Rows of Same Trace Eval Show up on Root Span (#50502)
Release 10.14.4
Features
Add option to pull images from op-hub arizeax (#50489)
Release 10.14.3
Bug Fixes
Fix image url encoding (#50356)
Release 10.14.2
Bug Fixes
Sessions page refetches until end of data (#50350)
Release 10.14.1
Bug Fixes
Improve online tasks custom gemini log
Release 10.14.0
Features
Make prompt optimization robust to json (#48295)
Playground - CMD Enter hot-key to run (#41563)
Rename experiments (#49467)
Bug Fixes
Issue running tasks on experiments (#49406)
Error when running prompt playground with custom model endpoint after loading in prompt (#49484)
Percent Volume in Slice Distributions Don't Exclude Actual Values of 0 (#42188)
Release 10.13.3
Bug Fixes
Error when running prompt playground with custom model endpoint after loading in prompt (#49484)
Release 10.13.2
Bug Fixes
Custom Model Endpoint fix for Gemini
Release 10.13.1
Bug Fixes
Calling Vertex AI from Custom Model Endpoint on Online Tasks does not work (#49846)
Release 10.13.0
Features
Highlights 07.2025
Support for service keys (#45752)
Improve Return Navigation from Detail Views for Custom Metrics, Monitors, Dashboards (#48508)
Export dataset and experiment as CSV (#39347)
Expose updateSpace mutation to end-users (#48870)
Support for additional Kyverno policies (#49033)
Support function calling for Gemini in playground (#46163)
Expose spanRecords as public (#49266)
Bug Fixes
Some eval templates use the same variable twice (#48666)
Test in Playground Pulls wrong model (#46716)
Error when running prompt playground with custom model endpoint after loading in prompt (#49484)
Wrong Image URL encoding with images in for traces (#47498)
Release 10.12.4
Bug Fixes
UI still attempts to send an email when reseting user (#49415)
Release 10.12.3
Bug Fixes
Add default AI models costs (#49356)
Release 10.12.2
Bug Fixes
Cosmetic logo changes (#49264)
Release 10.12.1
Features
Support for additional Kyverno policies (#49033)
Release 10.12.0
Features
Highlights 07.2025
Sankey agent flow (#46855)
Expand trace CSV download limit from 200 to 2000 (#44646)
Add Monitor Runtime Changes to Audit Log (#48802)
Support for Kubernetes 1.33 (#48848)
Prevent pods from being placed on karpenter spot instances (#48760)
Address some vulnerabilities (#48836)
Deprecate acr-secret and azurePrincipalId/azurePrincipalPassword (#49068)
Option to opt-out of creating roles in helm chart (#48474)
Option to opt-out of creating roles in the Operator (#49261)
Support mounting super secret with csi on Operator (#46893)
Support for additional Kyverno policies (#49033)
Bug Fixes
Tasks not getting called when running evaluators on experiments (#47463)
Issue with trace when looking at spans via embeddings page (#47222)
Unable to use model baseline on drift monitor even when set up on config page (#45435)
Release 10.11.1
Features
Support for Kubernetes 1.33 (#48848)
Bug Fixes
Address security / vulnerabilities (#48680) (#48836)
Release 10.11.0
Features
Highlights 06.2025
Alyx
Agent Viz
app-server pods should be placed on different nodes (#47575)
Breaking Changes in Helm Provider v3.0.0 (#48211)
Support for Kong ingress with single endpoint (#45335)
Add trace/session evals to UI (#47932)
Bug Fixes
Performance Tracing Slices Tab Color Scale Unexpectedly Changes for Same Metrics (#47342)
Datasets are missing (#48005)
Exporting Model Version Filter Not Working (#42862)
Playground UI is sluggish with loaded dataset (#46693)
Input variable values not being displayed in Prompt Playground (#47956)
Empty Annotators Name with Checkmarks in Labeling Queue UI (#46954)
The default value for kms additional keys is incorrect in on-prem terraform (#48366)
Fix gazette issue with node level role and AWS profile (#48551)
Feature Importance Sorting Doesn't Work (#37516)
Last updated
Was this helpful?