Feat/version 1.3.0

CHANGELOG.md

@@ -2,6 +2,12 @@
 
 All notable changes to `uipath_llm_client` (core package) will be documented in this file.
 
+## [1.3.0] - 2026-03-10
+
+### Version Bump
+- Stable release
+- Fixes
+
 ## [1.2.4] - 2026-02-26
 
 ### Refactor

packages/uipath_langchain_client/CHANGELOG.md

@@ -2,6 +2,12 @@
 
 All notable changes to `uipath_langchain_client` will be documented in this file.
 
+## [1.3.0] - 2026-03-10
+
+### Version Bump
+- Stable release
+- Fixes
+
 ## [1.2.7] - 2026-02-26
 
 ### Fix

packages/uipath_langchain_client/pyproject.toml

@@ -5,29 +5,29 @@ dynamic = ["version"]
 readme = "README.md"
 requires-python = ">=3.11"
 dependencies = [
-    "langchain>=1.2.7",
-    "uipath-llm-client>=1.2.4",
+    "langchain>=1.2.10",
+    "uipath-llm-client>=1.3.0",
 ]
 
 [project.optional-dependencies]
 openai = [
-    "langchain-openai>=1.1.7",
+    "langchain-openai>=1.1.11",
 ]
 aws = [
-    "langchain-aws>=1.2.1",
+    "langchain-aws>=1.4.0",
 ]
 google = [
-    "langchain-google-genai>=4.2.0",
+    "langchain-google-genai>=4.2.1",
 ]
 anthropic = [
-    "langchain-anthropic>=1.3.1",
-    "anthropic[bedrock,vertex]>=0.77.0",
+    "langchain-anthropic>=1.3.4",
+    "anthropic[bedrock,vertex]>=0.84.0",
 ]
 azure = [
-    "langchain-azure-ai>=1.0.0",
+    "langchain-azure-ai>=1.1.0",
 ]
 vertexai = [
-    "langchain-google-vertexai>=3.2.1",
+    "langchain-google-vertexai>=3.2.2",
 ]
 fireworks = [
     "langchain-fireworks>=1.1.0",

packages/uipath_langchain_client/src/uipath_langchain_client/__version__.py

@@ -1,3 +1,3 @@
 __title__ = "UiPath LangChain Client"
 __description__ = "A Python client for interacting with UiPath's LLM services via LangChain."
-__version__ = "1.2.7"
+__version__ = "1.3.0"

packages/uipath_langchain_client/src/uipath_langchain_client/base_client.py

@@ -102,7 +102,7 @@ class UiPathBaseLLMClient(BaseModel, ABC):
     )
 
     default_headers: Mapping[str, str] | None = Field(
-        default={
+        default_factory=lambda: {
             "X-UiPath-LLMGateway-TimeoutSeconds": "295",  # server side timeout, default is 10, maximum is 300
             "X-UiPath-LLMGateway-AllowFull4xxResponse": "true",  # allow full 4xx responses (default is false)
         },
@@ -210,7 +210,7 @@ def uipath_request(
     async def uipath_arequest(
         self,
         method: Literal["POST", "GET"] = "POST",
-        url: str = "/",
+        url: URL | str = "/",
         *,
         request_body: dict[str, Any] | None = None,
         **kwargs: Any,

packages/uipath_langchain_client/src/uipath_langchain_client/clients/azure/chat_models.py

@@ -1,46 +1,70 @@
 from typing import Self
 
-from pydantic import model_validator
+from httpx import URL, Request
+from pydantic import Field, model_validator
 
 from uipath_langchain_client.base_client import UiPathBaseChatModel
 from uipath_langchain_client.settings import UiPathAPIConfig
 
 try:
-    from azure.ai.inference import ChatCompletionsClient
-    from azure.ai.inference.aio import ChatCompletionsClient as ChatCompletionsClientAsync
-    from azure.core.credentials import AzureKeyCredential
-    from langchain_azure_ai.chat_models import AzureAIChatCompletionsModel
+    from azure.core.credentials import AzureKeyCredential, TokenCredential
+    from azure.core.credentials_async import AsyncTokenCredential
+    from langchain_azure_ai.chat_models import AzureAIOpenAIApiChatModel
+    from openai import AsyncOpenAI, OpenAI
 except ImportError as e:
     raise ImportError(
         "The 'azure' extra is required to use UiPathAzureAIChatCompletionsModel. "
         "Install it with: uv add uipath-langchain-client[azure]"
     ) from e
 
 
-class UiPathAzureAIChatCompletionsModel(UiPathBaseChatModel, AzureAIChatCompletionsModel):  # type: ignore[override]
+class UiPathAzureAIChatCompletionsModel(UiPathBaseChatModel, AzureAIOpenAIApiChatModel):  # type: ignore[override]
     api_config: UiPathAPIConfig = UiPathAPIConfig(
         api_type="completions",
         client_type="passthrough",
         vendor_type="azure",
-        freeze_base_url=True,
+        freeze_base_url=False,
     )
 
-    # Override fields to avoid errors when instantiating the class
-    endpoint: str | None = "PLACEHOLDER"
+    # Override fields to avoid env var lookup / validation errors at instantiation
+    endpoint: str | None = Field(default="PLACEHOLDER")
+    credential: str | AzureKeyCredential | TokenCredential | AsyncTokenCredential | None = Field(
+        default="PLACEHOLDER"
+    )
 
     @model_validator(mode="after")
     def setup_uipath_client(self) -> Self:
-        # TODO: finish implementation once we have a proper model in UiPath API
-        self._client = ChatCompletionsClient(
-            endpoint="PLACEHOLDER",
-            credential=AzureKeyCredential("PLACEHOLDER"),
-            model=self.model_name,
-            **self.client_kwargs,
+        base_url = str(self.uipath_sync_client.base_url).rstrip("/")
+
+        def fix_url_and_api_flavor_header(request: Request):
+            url_suffix = str(request.url).split(base_url)[-1]
+            if "responses" in url_suffix:
+                request.headers["X-UiPath-LlmGateway-ApiFlavor"] = "responses"
+            else:
+                request.headers["X-UiPath-LlmGateway-ApiFlavor"] = "chat-completions"
+            request.url = URL(base_url)
+
+        async def fix_url_and_api_flavor_header_async(request: Request):
+            url_suffix = str(request.url).split(base_url)[-1]
+            if "responses" in url_suffix:
+                request.headers["X-UiPath-LlmGateway-ApiFlavor"] = "responses"
+            else:
+                request.headers["X-UiPath-LlmGateway-ApiFlavor"] = "chat-completions"
+            request.url = URL(base_url)
+
+        self.uipath_sync_client.event_hooks["request"].append(fix_url_and_api_flavor_header)
+        self.uipath_async_client.event_hooks["request"].append(fix_url_and_api_flavor_header_async)
+
+        self.root_client = OpenAI(
+            api_key="PLACEHOLDER",
+            max_retries=0,  # handled by the UiPath client
+            http_client=self.uipath_sync_client,
         )
-        self._async_client = ChatCompletionsClientAsync(
-            endpoint="PLACEHOLDER",
-            credential=AzureKeyCredential("PLACEHOLDER"),
-            model=self.model_name,
-            **self.client_kwargs,
+        self.root_async_client = AsyncOpenAI(
+            api_key="PLACEHOLDER",
+            max_retries=0,  # handled by the UiPath client
+            http_client=self.uipath_async_client,
         )
+        self.client = self.root_client.chat.completions
+        self.async_client = self.root_async_client.chat.completions
         return self

packages/uipath_langchain_client/src/uipath_langchain_client/clients/azure/embeddings.py

@@ -1,46 +1,47 @@
 from typing import Self
 
-from pydantic import model_validator
+from pydantic import Field, model_validator
 
 from uipath_langchain_client.base_client import UiPathBaseEmbeddings
 from uipath_langchain_client.settings import UiPathAPIConfig
 
 try:
-    from azure.ai.inference import EmbeddingsClient
-    from azure.ai.inference.aio import EmbeddingsClient as EmbeddingsClientAsync
-    from langchain_azure_ai.embeddings import AzureAIEmbeddingsModel
+    from azure.core.credentials import AzureKeyCredential, TokenCredential
+    from azure.core.credentials_async import AsyncTokenCredential
+    from langchain_azure_ai.embeddings import AzureAIOpenAIApiEmbeddingsModel
+    from openai import AsyncOpenAI, OpenAI
 except ImportError as e:
     raise ImportError(
         "The 'azure' extra is required to use UiPathAzureAIEmbeddingsModel. "
         "Install it with: uv add uipath-langchain-client[azure]"
     ) from e
 
 
-class UiPathAzureAIEmbeddingsModel(UiPathBaseEmbeddings, AzureAIEmbeddingsModel):  # type: ignore[override]
+class UiPathAzureAIEmbeddingsModel(UiPathBaseEmbeddings, AzureAIOpenAIApiEmbeddingsModel):  # type: ignore[override]
     api_config: UiPathAPIConfig = UiPathAPIConfig(
         api_type="embeddings",
         client_type="passthrough",
         vendor_type="azure",
         freeze_base_url=True,
     )
 
-    # Override fields to avoid errors when instantiating the class
-    endpoint: str | None = "PLACEHOLDER"
-    credentials: str | None = "PLACEHOLDER"
+    # Override fields to avoid env var lookup / validation errors at instantiation
+    model: str = Field(default="", alias="model_name")
+    endpoint: str | None = Field(default="PLACEHOLDER")
+    credential: str | AzureKeyCredential | TokenCredential | AsyncTokenCredential | None = Field(
+        default="PLACEHOLDER"
+    )
 
     @model_validator(mode="after")
     def setup_uipath_client(self) -> Self:
-        # TODO: finish implementation once we have a proper model in UiPath API
-        self._client = EmbeddingsClient(
-            endpoint="PLACEHOLDER",
-            credentials="PLACEHOLDER",
-            model=self.model_name,
-            **self.client_kwargs,
-        )
-        self._async_client = EmbeddingsClientAsync(
-            endpoint="PLACEHOLDER",
-            credentials="PLACEHOLDER",
-            model=self.model_name,
-            **self.client_kwargs,
-        )
+        self.client = OpenAI(
+            api_key="PLACEHOLDER",
+            max_retries=0,  # handled by the UiPath client
+            http_client=self.uipath_sync_client,
+        ).embeddings
+        self.async_client = AsyncOpenAI(
+            api_key="PLACEHOLDER",
+            max_retries=0,  # handled by the UiPath client
+            http_client=self.uipath_async_client,
+        ).embeddings
         return self

packages/uipath_langchain_client/src/uipath_langchain_client/clients/normalized/chat_models.py

@@ -140,7 +140,7 @@ def _llm_type(self) -> str:
 
     @property
     def _default_params(self) -> dict[str, Any]:
-        """Get the default parameters for calling OpenAI API."""
+        """Get the default parameters for the normalized API request."""
         exclude_if_none = {
             "frequency_penalty": self.frequency_penalty,
             "presence_penalty": self.presence_penalty,
@@ -175,7 +175,7 @@ def _get_usage_metadata(self, json_data: dict[str, Any]) -> UsageMetadata:
                 cache_creation=json_data.get("cache_creation_input_tokens", 0),
             ),
             output_token_details=OutputTokenDetails(
-                audio=json_data.get("audio_tokens", 0),
+                audio=json_data.get("output_audio_tokens", 0),
                 reasoning=json_data.get("thoughts_tokens", 0),
             ),
         )
@@ -241,7 +241,15 @@ def _preprocess_request(
                         {
                             "id": tool_call["id"],
                             "name": tool_call["function"]["name"],
-                            "arguments": json.loads(tool_call["function"]["arguments"]),
+                            "arguments": (
+                                tool_call["function"]["arguments"]
+                                if isinstance(tool_call["function"]["arguments"], dict)
+                                else (
+                                    json.loads(tool_call["function"]["arguments"])
+                                    if tool_call["function"]["arguments"]
+                                    else {}
+                                )
+                            ),
                         }
                         for tool_call in converted_message["tool_calls"]
                     ]

packages/uipath_langchain_client/src/uipath_langchain_client/utils.py

@@ -0,0 +1,33 @@
+from uipath.llm_client.utils.exceptions import (
+    UiPathAPIError,
+    UiPathAuthenticationError,
+    UiPathBadRequestError,
+    UiPathConflictError,
+    UiPathGatewayTimeoutError,
+    UiPathInternalServerError,
+    UiPathNotFoundError,
+    UiPathPermissionDeniedError,
+    UiPathRateLimitError,
+    UiPathRequestTooLargeError,
+    UiPathServiceUnavailableError,
+    UiPathTooManyRequestsError,
+    UiPathUnprocessableEntityError,
+)
+from uipath.llm_client.utils.retry import RetryConfig
+
+__all__ = [
+    "RetryConfig",
+    "UiPathAPIError",
+    "UiPathBadRequestError",
+    "UiPathAuthenticationError",
+    "UiPathPermissionDeniedError",
+    "UiPathNotFoundError",
+    "UiPathConflictError",
+    "UiPathRequestTooLargeError",
+    "UiPathUnprocessableEntityError",
+    "UiPathRateLimitError",
+    "UiPathInternalServerError",
+    "UiPathServiceUnavailableError",
+    "UiPathGatewayTimeoutError",
+    "UiPathTooManyRequestsError",
+]

pyproject.toml

@@ -6,10 +6,10 @@ readme = "README.md"
 requires-python = ">=3.11"
 dependencies = [
     "httpx>=0.28.1",
-    "tenacity>=9.1.2",
+    "tenacity>=9.1.4",
     "pydantic>=2.12.5",
-    "pydantic-settings>=2.12.0",
-    "uipath>=2.5.17",
+    "pydantic-settings>=2.13.1",
+    "uipath>=2.10.11",
 ]
 
 authors = [
@@ -19,26 +19,26 @@ authors = [
 
 [project.optional-dependencies]
 openai = [
-    "openai>=2.15.0",
+    "openai>=2.24.0",
 ]
 google = [
-    "google-genai>=1.59.0",
+    "google-genai>=1.65.0",
 ]
 anthropic = [
-    "anthropic>=0.76.0",
+    "anthropic>=0.84.0",
 ]
 all = [
     "uipath-llm-client[openai,google,anthropic]",
 ]
 
 [dependency-groups]
 dev = [
-    "langchain-tests>=1.1.2",
-    "pytest>=8.4.2",
+    "langchain-tests>=1.1.5",
+    "pytest>=9.0.2",
     "pytest-asyncio>=1.3.0",
     "pytest-recording>=0.13.4",
-    "pyright>=1.1.407",
-    "ruff>=0.14.9",
+    "pyright>=1.1.408",
+    "ruff>=0.15.3",
     "uipath-llm-client[all]",
     "uipath_langchain_client[all]",
 ]

src/uipath/llm_client/__version__.py

@@ -1,3 +1,3 @@
-__titile__ = "UiPath LLM Client"
+__title__ = "UiPath LLM Client"
 __description__ = "A Python client for interacting with UiPath's LLM services."
-__version__ = "1.2.4"
+__version__ = "1.3.0"

src/uipath/llm_client/settings/llmgateway/auth.py

@@ -30,8 +30,8 @@ def get_llmgw_token_header(
     ) -> str:
         """Retrieve a new access token from the LLM Gateway identity endpoint."""
         url_get_token = f"{self.settings.base_url}/{LLMGatewayEndpoints.IDENTITY_ENDPOINT.value}"
-        assert self.settings.client_id is not None
-        assert self.settings.client_secret is not None
+        if self.settings.client_id is None or self.settings.client_secret is None:
+            raise ValueError("client_id and client_secret are required for S2S authentication")
         token_credentials = dict(
             client_id=self.settings.client_id.get_secret_value(),
             client_secret=self.settings.client_secret.get_secret_value(),
@@ -40,11 +40,15 @@ def get_llmgw_token_header(
         with Client() as http_client:
             response = http_client.post(url_get_token, data=token_credentials)
             if response.is_client_error:
+                try:
+                    body = response.json()
+                except Exception:
+                    body = response.text
                 raise UiPathAuthenticationError(
                     message="Failed to authenticate with LLM Gateway, invalid credentials",
                     request=response.request,
                     response=response,
-                    body=response.json(),
+                    body=body,
                 )
             elif response.is_error:
                 raise UiPathAPIError.from_response(response)