You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Copy file name to clipboardExpand all lines: src/langsmith/self-hosted-changelog.mdx
+29Lines changed: 29 additions & 0 deletions
Display the source diff
Display the rich diff
Original file line number
Diff line number
Diff line change
@@ -34,6 +34,35 @@ rss: true
34
34
**Download the Helm chart:**[`langsmith-observability-0.2.0.tgz`](https://github.com/langchain-ai/helm/releases/download/langsmith-observability-0.2.0/langsmith-observability-0.2.0.tgz)
35
35
</Update>
36
36
37
+
<Updatelabel="2026-03-07"tags={["self-hosted"]}>
38
+
## langsmith-observability-0.2.0
39
+
40
+
- Fixed an insights agent bug that was not persisting categories.
41
+
- Improved the UI for async sandbox creation with status polling.
42
+
- Added JSON/YAML syntax highlighting to the experiment comparison feature.
43
+
- Improved thread trace open behavior and removed the expand button in the frontend.
44
+
- Added a new sandbox endpoint to smith-go.
45
+
- Supported OpenAI-compatible endpoints for smith-polly.
46
+
- Fixed the timeout issue with bulk exports stuck in CREATED status.
47
+
- Allowed service identity access to create repo endpoint.
48
+
- Recorded hub prompt commit in experiment session metadata in the playground.
49
+
- Improved authorization for the `/sessions` shadow query.
50
+
- Added new ABAC backend deployments (v2) for better security management.
51
+
- Implemented projects and runs write permissions in the UI.
52
+
- Added support for GPT-5.4 and GPT-5.4-pro models, including pricing and playground integration.
53
+
- Made GPT-5.4 the default model for the OpenAI playground.
54
+
- Increased the maximum number of tags displayed in the RunTags component.
55
+
- Fixed an agent builder issue where runs were rejected when limits changed.
56
+
- Resolved float errors in the /sessions go endpoint.
57
+
- Ensured Redis cache returns fetched value when cache SET fails.
58
+
- Supported AWS IAM roles for the agent builder, polly, and insights integration.
59
+
- Improved the custom chart CRUD interface on the frontend.
60
+
- Added the ability to filter and group experiments by prompt in the experiments table.
61
+
- Updated inbox counts and thread fetching logic in the agent builder.
62
+
63
+
**Download the Helm chart:**[`langsmith-observability-0.2.0.tgz`](https://github.com/langchain-ai/helm/releases/download/langsmith-observability-0.2.0/langsmith-observability-0.2.0.tgz)
0 commit comments