$ the-wire · showcase
LangChain Azure AD auth rides OpenAI 3.8.0
By RepoJournal · Filed · About AI Agents
LangChain upgraded its OpenAI SDK lock to 3.8.0, restoring Azure AD token provider support across chat, embeddings, and completions without API-key conflicts.
LangChain pinned the OpenAI Python SDK to 3.8.0 while keeping the existing lower bound, and adjusted Azure AD handling to match the SDK's credential validation [1]. This means AzureChatOpenAI, Azure embeddings, and Azure completions again accept Azure AD token providers without tripping over API-key validation, including async token providers.
OpenHands fixed a serialization bug where buildConfiguredOpenHandsAgentSettings stripped base_url from subscription LLM configs; that could leave runtime configs missing the Azure endpoint (https://chatgpt.com/backend-api/codex) [2]. The fix applies on the client side, agent-canvas, rather than patching the SDK.
LangChain also cleaned up stale Args/Raises entries in FileCallbackHandler._write and ChatGeneration.set_text documentation [3].
The changes are routine; the Azure AD fix matters if your deployments rely on token-based auth with the latest OpenAI SDK [1].
Action items
- → Verify Azure AD token auth still works after upgrading LangChain's OpenAI SDK lock to 3.8.0 langchain-ai/langchain [monitor]
- → Upgrade OpenHands to pick up the base_url preservation fix if you use subscription LLM configs OpenHands/OpenHands [plan]
References
- [1] fix(openai): support Azure AD auth with OpenAI 3.8 ↗ langchain-ai/langchain
- [2] fix: preserve base_url for subscription LLM configs ↗ OpenHands/OpenHands
- [3] docs(core): remove stale Args/Raises entries from FileCallbackHandler._write and ChatGeneration.set_text (#40211) ↗ langchain-ai/langchain