feat(agents): add OVHcloud chat completions and update to OpenAI-compatible endpoint

libs/community/langchain_community/chat_models/__init__.py

@@ -140,6 +140,9 @@
     from langchain_community.chat_models.ollama import (
         ChatOllama,
     )
+    from langchain_community.chat_models.ovhcloud import (
+        ChatOVHcloud,
+    )
     from langchain_community.chat_models.openai import (
         ChatOpenAI,
     )
@@ -204,6 +207,7 @@
     "ChatCohere",
     "ChatCoze",
     "ChatOctoAI",
+    "ChatOVHcloud",
     "ChatDatabricks",
     "ChatDeepInfra",
     "ChatEdenAI",
@@ -290,6 +294,7 @@
     "ChatMlflow": "langchain_community.chat_models.mlflow",
     "ChatNebula": "langchain_community.chat_models.symblai_nebula",
     "ChatOctoAI": "langchain_community.chat_models.octoai",
+    "ChatOVHcloud": "langchain_community.chat_models.ovhcloud",
     "ChatOCIGenAI": "langchain_community.chat_models.oci_generative_ai",
     "ChatOCIModelDeployment": "langchain_community.chat_models.oci_data_science",
     "ChatOCIModelDeploymentVLLM": "langchain_community.chat_models.oci_data_science",

libs/community/langchain_community/chat_models/ovhcloud.py

@@ -0,0 +1,161 @@
+"""OVHcloud AI Endpoints chat wrapper. Relies heavily on ChatOpenAI."""
+
+from typing import (
+    Any,
+    Callable,
+    Dict,
+    Literal,
+    Optional,
+    Sequence,
+    Type,
+    Union,
+)
+
+from langchain_core.language_models import LanguageModelInput
+from langchain_core.messages import AIMessage
+from langchain_core.runnables import Runnable
+from langchain_core.tools import BaseTool
+from langchain_core.utils import convert_to_secret_str, get_from_dict_or_env, pre_init
+from langchain_core.utils.function_calling import convert_to_openai_tool
+from pydantic import Field, SecretStr
+
+from langchain_community.chat_models.openai import ChatOpenAI
+from langchain_community.utils.openai import is_openai_v1
+
+DEFAULT_API_BASE = "https://oai.endpoints.kepler.ai.cloud.ovh.net/v1"
+DEFAULT_MODEL = "gpt-oss-120b"
+
+
+class ChatOVHcloud(ChatOpenAI):
+    """OVHcloud AI Endpoints Chat large language models.
+
+    See https://www.ovhcloud.com/en/public-cloud/ai-endpoints/catalog/ for information about OVHcloud AI Endpoints.
+
+    To use, you should have the ``openai`` python package installed and the
+    environment variable ``OVHCLOUD_API_TOKEN`` set with your API token.
+    Alternatively, you can use the ovhcloud_api_token keyword argument.
+
+    Any parameters that are valid to be passed to the `openai.create` call can be passed
+    in, even if not explicitly saved on this class.
+
+    Example:
+        .. code-block:: python
+
+            from langchain_community.chat_models import ChatOVHcloud
+            chat = ChatOVHcloud(model_name="gpt-oss-120b")
+    """
+
+    ovhcloud_api_base: str = Field(default=DEFAULT_API_BASE)
+    ovhcloud_api_token: SecretStr = Field(default=SecretStr(""), alias="api_key")
+    model_name: str = Field(default=DEFAULT_MODEL, alias="model")
+
+    @property
+    def _llm_type(self) -> str:
+        """Return type of chat model."""
+        return "ovhcloud-chat"
+
+    @property
+    def lc_secrets(self) -> Dict[str, str]:
+        return {"ovhcloud_api_token": "OVHCLOUD_API_TOKEN"}
+
+    @classmethod
+    def is_lc_serializable(cls) -> bool:
+        return False
+
+    @pre_init
+    def validate_environment(cls, values: Dict) -> Dict:
+        """Validate that api key and python package exists in environment."""
+        values["ovhcloud_api_base"] = get_from_dict_or_env(
+            values,
+            "ovhcloud_api_base",
+            "OVHCLOUD_API_BASE",
+            default=DEFAULT_API_BASE,
+        )
+        values["ovhcloud_api_token"] = convert_to_secret_str(
+            get_from_dict_or_env(
+                values, "ovhcloud_api_token", "OVHCLOUD_API_TOKEN"
+            )
+        )
+        values["model_name"] = get_from_dict_or_env(
+            values,
+            "model_name",
+            "OVHCLOUD_MODEL_NAME",
+            default=DEFAULT_MODEL,
+        )
+
+        try:
+            import openai
+
+            if is_openai_v1():
+                client_params = {
+                    "api_key": values["ovhcloud_api_token"].get_secret_value(),
+                    "base_url": values["ovhcloud_api_base"],
+                }
+                if not values.get("client"):
+                    values["client"] = openai.OpenAI(**client_params).chat.completions
+                if not values.get("async_client"):
+                    values["async_client"] = openai.AsyncOpenAI(
+                        **client_params
+                    ).chat.completions
+            else:
+                values["openai_api_base"] = values["ovhcloud_api_base"]
+                values["openai_api_key"] = values["ovhcloud_api_token"].get_secret_value()
+                values["client"] = openai.ChatCompletion
+        except ImportError:
+            raise ImportError(
+                "Could not import openai python package. "
+                "Please install it with `pip install openai`."
+            )
+
+        return values
+
+    def bind_tools(
+        self,
+        tools: Sequence[Union[Dict[str, Any], Type, Callable, BaseTool]],
+        *,
+        tool_choice: Optional[
+            Union[dict, str, Literal["auto", "none", "required", "any"], bool]
+        ] = None,
+        strict: Optional[bool] = None,
+        **kwargs: Any,
+    ) -> Runnable[LanguageModelInput, AIMessage]:
+        """Imitating bind_tool method from langchain_openai.ChatOpenAI"""
+
+        formatted_tools = [
+            convert_to_openai_tool(tool, strict=strict) for tool in tools
+        ]
+        if tool_choice:
+            if isinstance(tool_choice, str):
+                # tool_choice is a tool/function name
+                if tool_choice not in ("auto", "none", "any", "required"):
+                    tool_choice = {
+                        "type": "function",
+                        "function": {"name": tool_choice},
+                    }
+                # 'any' is not natively supported by OpenAI API.
+                # We support 'any' since other models use this instead of 'required'.
+                if tool_choice == "any":
+                    tool_choice = "required"
+            elif isinstance(tool_choice, bool):
+                tool_choice = "required"
+            elif isinstance(tool_choice, dict):
+                tool_names = [
+                    formatted_tool["function"]["name"]
+                    for formatted_tool in formatted_tools
+                ]
+                if not any(
+                    tool_name == tool_choice["function"]["name"]
+                    for tool_name in tool_names
+                ):
+                    raise ValueError(
+                        f"Tool choice {tool_choice} was specified, but the only "
+                        f"provided tools were {tool_names}."
+                    )
+            else:
+                raise ValueError(
+                    f"Unrecognized tool_choice type. Expected str, bool or dict. "
+                    f"Received: {tool_choice}"
+                )
+            kwargs["tool_choice"] = tool_choice
+        return super().bind(tools=formatted_tools, **kwargs)
+

libs/community/langchain_community/embeddings/ovhcloud.py

@@ -1,9 +1,9 @@
-import json
 import logging
 import time
 from typing import Any, List
 
 import requests
+from langchain_core._api.deprecation import deprecated
 from langchain_core.embeddings import Embeddings
 from pydantic import BaseModel, ConfigDict
 
@@ -21,9 +21,16 @@ class OVHCloudEmbeddings(BaseModel, Embeddings):
     """ OVHcloud AI Endpoints model name for embeddings generation"""
     model_name: str = ""
 
-    """ OVHcloud AI Endpoints region"""
+    @deprecated(
+        since="0.4.2",
+        removal="1.0",
+    )
+    """ OVHcloud AI Endpoints region (deprecated, kept for backward compatibility)"""
     region: str = "kepler"
 
+    """ OVHcloud AI Endpoints base URL"""
+    base_url: str = "https://oai.endpoints.kepler.ai.cloud.ovh.net/v1"
+
     model_config = ConfigDict(extra="forbid", protected_namespaces=())
 
     def __init__(self, **kwargs: Any):
@@ -32,8 +39,6 @@ def __init__(self, **kwargs: Any):
             raise ValueError("Access token is required for OVHCloud embeddings.")
         if self.model_name == "":
             raise ValueError("Model name is required for OVHCloud embeddings.")
-        if self.region == "":
-            raise ValueError("Region is required for OVHCloud embeddings.")
 
     def _generate_embedding(self, text: str) -> List[float]:
         """Generate embeddings from OVHCLOUD AIE.
@@ -42,8 +47,7 @@ def _generate_embedding(self, text: str) -> List[float]:
         Returns:
             List[float]: Embeddings for the text.
         """
-
-        return self._send_request_to_ai_endpoints("text/plain", text, "text2vec")
+        return self._send_request_to_ai_endpoints([text])[0]
 
     def embed_documents(self, texts: List[str]) -> List[List[float]]:
         """Embed a list of documents.
@@ -54,10 +58,7 @@ def embed_documents(self, texts: List[str]) -> List[List[float]]:
            List[List[float]]: List of embeddings, one for each input text.
 
         """
-
-        return self._send_request_to_ai_endpoints(
-            "application/json", json.dumps(texts), "batch_text2vec"
-        )
+        return self._send_request_to_ai_endpoints(texts)
 
     def embed_query(self, text: str) -> List[float]:
         """Embed a single query text.
@@ -68,29 +69,31 @@ def embed_query(self, text: str) -> List[float]:
         """
         return self._generate_embedding(text)
 
-    def _send_request_to_ai_endpoints(
-        self, contentType: str, payload: str, route: str
-    ) -> Any:
-        """Send a HTTPS request to OVHcloud AI Endpoints
+    def _send_request_to_ai_endpoints(self, texts: List[str]) -> List[List[float]]:
+        """Send a HTTPS request to OVHcloud AI Endpoints using OpenAI-compatible API
         Args:
-            contentType (str): The content type of the request, application/json or text/plain.
-            payload (str): The payload of the request.
-            route (str): The route of the request, batch_text2vec or text2vec.
-        """  # noqa: E501
+            texts (List[str]): The list of texts to embed.
+        Returns:
+            List[List[float]]: List of embeddings, one for each input text.
+        """
         headers = {
-            "content-type": contentType,
+            "Content-Type": "application/json",
             "Authorization": f"Bearer {self.access_token}",
         }
 
+        # Prepare request body in OpenAI format
+        # OpenAI API accepts both string and array, but we always use array for consistency
+        payload = {
+            "model": self.model_name,
+            "input": texts,
+        }
+
         session = requests.session()
         while True:
             response = session.post(
-                (
-                    f"https://{self.model_name}.endpoints.{self.region}"
-                    f".ai.cloud.ovh.net/api/{route}"
-                ),
+                f"{self.base_url}/embeddings",
                 headers=headers,
-                data=payload,
+                json=payload,
             )
             if response.status_code != 200:
                 if response.status_code == 429:
@@ -112,4 +115,8 @@ def _send_request_to_ai_endpoints(
                         status_code=response.status_code, text=response.text
                     )
                 )
-            return response.json()
+            # Parse OpenAI-compatible response format
+            response_data = response.json()
+            # OpenAI format: {"data": [{"embedding": [...]}, ...]}
+            embeddings = [item["embedding"] for item in response_data["data"]]
+            return embeddings

libs/community/tests/unit_tests/chat_models/test_imports.py

@@ -63,6 +63,7 @@
     "QianfanChatEndpoint",
     "VolcEngineMaasChat",
     "ChatOctoAI",
+    "ChatOVHcloud",
     "ChatSnowflakeCortex",
     "ChatYi",
 ]

libs/community/tests/unit_tests/chat_models/test_ovhcloud.py

@@ -0,0 +1,52 @@
+import pytest
+from pydantic import SecretStr, ValidationError
+
+from langchain_community.chat_models.ovhcloud import ChatOVHcloud
+
+DEFAULT_API_BASE = "https://oai.endpoints.kepler.ai.cloud.ovh.net/v1"
+DEFAULT_MODEL = "gpt-oss-120b"
+
+
+@pytest.mark.requires("openai")
+def test__default_ovhcloud_api_base() -> None:
+    chat = ChatOVHcloud(ovhcloud_api_token=SecretStr("test_token"))  # type: ignore[call-arg]
+    assert chat.ovhcloud_api_base == DEFAULT_API_BASE
+
+
+@pytest.mark.requires("openai")
+def test__default_ovhcloud_api_token() -> None:
+    chat = ChatOVHcloud(ovhcloud_api_token=SecretStr("test_token"))  # type: ignore[call-arg]
+    assert chat.ovhcloud_api_token.get_secret_value() == "test_token"
+
+
+@pytest.mark.requires("openai")
+def test__default_model_name() -> None:
+    chat = ChatOVHcloud(ovhcloud_api_token=SecretStr("test_token"))  # type: ignore[call-arg]
+    assert chat.model_name == DEFAULT_MODEL
+
+
+@pytest.mark.requires("openai")
+def test__field_aliases() -> None:
+    chat = ChatOVHcloud(ovhcloud_api_token=SecretStr("test_token"), model="custom-model")  # type: ignore[call-arg]
+    assert chat.model_name == "custom-model"
+    assert chat.ovhcloud_api_token.get_secret_value() == "test_token"
+
+
+@pytest.mark.requires("openai")
+def test__missing_ovhcloud_api_token() -> None:
+    with pytest.raises(ValidationError) as e:
+        ChatOVHcloud()
+    assert "Did not find ovhcloud_api_token" in str(e)
+
+
+@pytest.mark.requires("openai")
+def test__all_fields_provided() -> None:
+    chat = ChatOVHcloud(  # type: ignore[call-arg]
+        ovhcloud_api_token=SecretStr("test_token"),
+        model="custom-model",
+        ovhcloud_api_base="https://custom.api/base/",
+    )
+    assert chat.ovhcloud_api_base == "https://custom.api/base/"
+    assert chat.ovhcloud_api_token.get_secret_value() == "test_token"
+    assert chat.model_name == "custom-model"
+