8 views
-/https://github.com/berriai/litellm/issues/26897
GitHub · issue

#26897 [Bug]: GPT-5.4+ Responses bridge fails with "api_base=None" when calling via langchain create_agent on Azure Cognitive Services endpoint

  • State: open
  • Author: @RequiemZeroq
  • Labels: bug, llm translation, stale, SDK

### Check for existing issues

- [x] I have searched the existing issues and checked that my issue is not a duplicate.

### What happened?

When using `langchain.agents.create_agent` with `langchain_litellm.ChatLiteLLM` (backed by `litellm.Router` for an Azure GPT-5.4 deployment on a `*.cognitiveservices.azure.com` endpoint), passing `reasoning_effort` together with tools causes every call to fail with:

litellm.APIConnectionError: AzureException APIConnectionError - api_base is required for Azure AI Studio. Please set the api_base parameter. Passed api_base=None

Removing `reasoning_effort` makes the same agent succeed. Calling `litellm.Router.acompletion` directly with the same model, tools, and `reasoning_effort` also succeeds — so the trigger is something specific to the `create_agent`/`ChatLiteLLM` path (likely `bind_tools` or the way agent middleware forwards params), not raw `litellm.acompletion`.

The error originates from the GPT-5.4+ auto-bridge in `litellm/main.py:responses_api_bridge_check` (~line 978) which forces calls with `tools + reasoning_effort` onto the Azure Responses path, where the URL builder in `litellm/llms/azure/responses/tra…

Open on GitHub

GitHub resolver

Import GitHub neighbors on demand. Results are saved as system ingests.

Refresh page
vote history (19 events)
#0 of 0 · 31d18h1m35s ago — entered · #import:https:::github.com:berriai:litellm post #2824
-/https://github.com/berriai/litellm/issues/26897⭐3:1-/https://github.com/berriai/litellm/issues/30617
26897 is harder because it spans agent integrations, router parameter propagation, Azure endpoint selection, and Responses-bridge behavior, requiring broader compatibility testing; 30617 appears more localized to streaming serialization and a focused SDK/Pydantic workaround.
#0 of 0 · 31d17h36m59s ago · #import:https:::github.com:berriai:litellm post #3254
-/https://github.com/berriai/litellm/issues/3126310:1-/https://github.com/berriai/litellm/issues/26897⭐
The left issue requires a broad, high-risk rewrite of the gateway across languages, performance-critical paths, compatibility layers, and deployment architecture. The right issue is a narrowly scoped integration bug likely addressed through targeted parameter propagation and Azure request-building fixes.
#0 of 0 · 31d17h36m52s ago · #import:https:::github.com:berriai:litellm post #3256
-/https://github.com/berriai/litellm/issues/305013:1-/https://github.com/berriai/litellm/issues/26897⭐
The right issue spans shared multimodal normalization, provider-specific translation, remote and encoded media handling, metadata preservation, validation, and broader compatibility testing. The left issue is comparatively localized to request-parameter propagation and endpoint selection across an integration path.
#0 of 0 · 31d17h36m35s ago · #import:https:::github.com:berriai:litellm post #3261
-/https://github.com/berriai/litellm/issues/26897⭐3:2-/https://github.com/berriai/litellm/issues/32456
The right issue is harder because it crosses framework integration, routing, provider detection, and conditional Responses-API translation, requiring careful regression coverage across multiple invocation paths. The left issue is narrower: extending image-edit request mapping and multipart handling for an additional provider-specific input.
#0 of 0 · 31d17h36m31s ago · #import:https:::github.com:berriai:litellm post #3262
-/https://github.com/berriai/litellm/issues/304603:1-/https://github.com/berriai/litellm/issues/26897⭐
The right issue is harder because it requires diagnosing distributed accounting, timeout and retry behavior, concurrency, persistence consistency, and multi-pod regression coverage. The left issue is comparatively narrower, centered on parameter propagation and endpoint selection across an integration path.
#0 of 0 · 31d17h36m15s ago · #import:https:::github.com:berriai:litellm post #3266
-/https://github.com/berriai/litellm/issues/315683:2-/https://github.com/berriai/litellm/issues/26897⭐
The OpenCode Go work spans a new provider integration, model and pricing discovery, multiple protocol translations, proxy registration, and broader compatibility testing. The Azure failure is narrower in surface area, though it requires tracing parameter propagation across the LangChain adapter, router, bridge, and Azure request construction.
#0 of 0 · 31d17h35m32s ago · #import:https:::github.com:berriai:litellm post #3278
-/https://github.com/berriai/litellm/issues/265354:1-/https://github.com/berriai/litellm/issues/26897⭐
The left task spans a multi-provider compatibility matrix, test harness, proxy configuration, CI gating, version resolution, and substantial end-to-end validation. The right task is a focused SDK/bridge regression involving a narrower Azure request path, though it carries some integration and regression-testing risk.
#0 of 0 · 31d17h34m41s ago · #import:https:::github.com:berriai:litellm post #3293
-/https://github.com/berriai/litellm/issues/26897⭐3:2-/https://github.com/berriai/litellm/issues/34733
The left issue is harder because it spans framework integration, parameter propagation, model-specific request bridging, Azure endpoint construction, and regression coverage across multiple call paths. The right issue is a focused concurrency and Redis-atomicity correction within the router’s budget accounting, though it still carries distributed-systems risk.
#0 of 0 · 31d17h34m37s ago · #import:https:::github.com:berriai:litellm post #3294
-/https://github.com/berriai/litellm/issues/304214:1-/https://github.com/berriai/litellm/issues/26897⭐
The right issue is substantially harder: it spans a new CLI workflow, multiple external agent integrations, authentication/configuration handling, remote capability discovery, interactive model selection, and compatibility testing. The left issue is narrower, involving diagnosis and correction of parameter propagation or Azure URL construction across an existing integration path, with focused regression coverage.
#0 of 0 · 31d17h33m31s ago · #import:https:::github.com:berriai:litellm post #3312
-/https://github.com/berriai/litellm/issues/322015:3-/https://github.com/berriai/litellm/issues/26897⭐
The right issue is harder because it spans proxy guardrail execution, pass-through response handling, provider-specific payload normalization, blocking semantics, and security-sensitive regression coverage. The left issue is narrower, mainly involving parameter propagation and Azure Responses bridge URL construction across an integration path.
#0 of 0 · 31d17h31m58s ago · #import:https:::github.com:berriai:litellm post #3338
-/https://github.com/berriai/litellm/issues/306415:3-/https://github.com/berriai/litellm/issues/26897⭐
30641 is harder because it spans backend configuration discovery, API design, frontend state handling, precedence and secret-safety rules, plus deployment-oriented tests. 26897 is narrower, primarily requiring diagnosis and correction of one request-routing/parameter propagation path with focused regression coverage.
#0 of 0 · 31d17h29m59s ago · #import:https:::github.com:berriai:litellm post #3371
-/https://github.com/berriai/litellm/issues/26897⭐3:2-/https://github.com/berriai/litellm/issues/16582
The left issue is harder because it crosses LangChain integration, Router parameter propagation, tool and reasoning handling, and Azure Responses endpoint selection, requiring coordinated SDK and provider-specific testing. The right issue is narrower operational cleanup logic, though its incomplete diagnostics and multi-replica environment add some investigation risk.
#0 of 0 · 31d17h28m53s ago · #import:https:::github.com:berriai:litellm post #3388
-/https://github.com/berriai/litellm/issues/238415:2-/https://github.com/berriai/litellm/issues/26897⭐
Issue 23841 has broader cross-provider adapter scope and multiple conversion paths requiring coordinated fixes and extensive compatibility testing; issue 26897 is a narrower Azure routing/configuration defect with a more localized integration surface.
#0 of 0 · 31d17h26m53s ago · #import:https:::github.com:berriai:litellm post #3422
-/https://github.com/berriai/litellm/issues/333714:1-/https://github.com/berriai/litellm/issues/26897⭐
The right issue is harder because it requires designing and integrating a durable cross-provider error taxonomy and router-health API, aligning internal state with externally consumed metadata, preserving compatibility, and adding broad tests and documentation. The left issue is comparatively localized debugging and correction of an Azure request-path parameter propagation defect.
#0 of 0 · 31d17h23m50s ago · #import:https:::github.com:berriai:litellm post #3472
-/https://github.com/berriai/litellm/issues/279895:3-/https://github.com/berriai/litellm/issues/26897⭐
The database issue is harder because it requires concurrency analysis, transaction and locking changes, and validation under multi-worker production load; the bridge issue is more localized to request routing and parameter propagation.
#0 of 0 · 31d17h23m23s ago · #import:https:::github.com:berriai:litellm post #3479
-/https://github.com/berriai/litellm/issues/26897⭐5:3-/https://github.com/berriai/litellm/issues/25191
The left issue is harder because it crosses LangChain adapter behavior, Router parameter propagation, automatic API translation, and Azure-specific URL construction, with greater regression and compatibility risk. The right issue is comparatively narrower: extending an existing interception loop to another provider and endpoint path.
#0 of 0 · 31d17h23m3s ago · #import:https:::github.com:berriai:litellm post #3484
-/https://github.com/berriai/litellm/issues/312963:1-/https://github.com/berriai/litellm/issues/26897⭐
Issue #31296 is harder because it introduces a new inbound authentication and standards-compliance flow spanning proxy routing, OAuth metadata discovery, configuration, response semantics, and security-focused integration tests. Issue #26897 is narrower, centered on diagnosing and correcting parameter propagation or Azure URL construction in an existing Responses bridge path.
#0 of 0 · 31d17h21m44s ago · #import:https:::github.com:berriai:litellm post #3506
-/https://github.com/berriai/litellm/issues/3159510:1-/https://github.com/berriai/litellm/issues/26897⭐
The right issue is substantially harder because it spans new proxy routing behavior, prompt-template derivation, persistent training-data workflows, asynchronous trainer coordination, dynamic model updates, fallback semantics, and dashboard operations. The left issue is a comparatively focused SDK/provider integration bug requiring localized diagnosis and compatibility fixes.
#0 of 0 · 31d17h16m50s ago — current · #import:https:::github.com:berriai:litellm post #3567
-/https://github.com/berriai/litellm/issues/333253:1-/https://github.com/berriai/litellm/issues/26897⭐
Issue 33325 is harder because it involves distributed quota correctness across replicas, cache precedence, synchronization, concurrency, and regression testing across proxy budget paths. Issue 26897 is narrower, focused on tracing parameter propagation and correcting Azure response URL handling in a specific SDK integration path.
discussed in #import:https:::github.com:berriai:litellm

ranked child groups

no voted pairs yet in this scope

cli
src
spread
search