langchain-google-genai 2.1.7__tar.gz → 2.1.9__tar.gz

{langchain_google_genai-2.1.7 → langchain_google_genai-2.1.9}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: langchain-google-genai
-Version: 2.1.7
+Version: 2.1.9
 Summary: An integration package connecting Google's genai package and LangChain
 Home-page: https://github.com/langchain-ai/langchain-google
 License: MIT
@@ -29,16 +29,20 @@ This package enables seamless access to Google Gemini's chat, vision, embeddings
 
 ## Table of Contents
 
-- [Overview](#overview)
-- [Installation](#installation)
-- [Quickstart](#quickstart)
-- [Chat Models](#chat-models)
-  - [Multimodal Inputs](#multimodal-inputs)
-  - [Multimodal Outputs](#multimodal-outputs)
-  - [Multimodal Outputs in Chains](#multimodal-outputs-in-chains)
-  - [Thinking Support](#thinking-support)
-- [Embeddings](#embeddings)
-- [Semantic Retrieval (RAG)](#semantic-retrieval-rag)
+- [langchain-google-genai](#langchain-google-genai)
+  - [Table of Contents](#table-of-contents)
+  - [Overview](#overview)
+  - [Installation](#installation)
+  - [Quickstart](#quickstart)
+  - [Chat Models](#chat-models)
+    - [Multimodal Inputs](#multimodal-inputs)
+    - [Multimodal Outputs](#multimodal-outputs)
+    - [Audio Output](#audio-output)
+    - [Multimodal Outputs in Chains](#multimodal-outputs-in-chains)
+    - [Thinking Support](#thinking-support)
+  - [Embeddings](#embeddings)
+  - [Semantic Retrieval (RAG)](#semantic-retrieval-rag)
+  - [Resources](#resources)
 
 ---
 
@@ -109,9 +113,9 @@ print(response.content)
 
 ✅ `image_url` can be:
 
-* A public image URL
-* A Google Cloud Storage path (`gcs://...`)
-* A base64-encoded image (e.g., `data:image/png;base64,...`)
+- A public image URL
+- A Google Cloud Storage path (`gcs://...`)
+- A base64-encoded image (e.g., `data:image/png;base64,...`)
 
 ---
 
@@ -208,7 +212,7 @@ You can use Gemini embeddings in LangChain:
 ```python
 from langchain_google_genai import GoogleGenerativeAIEmbeddings
 
-embeddings = GoogleGenerativeAIEmbeddings(model="models/embedding-001")
+embeddings = GoogleGenerativeAIEmbeddings(model="models/gemini-embedding-001")
 vector = embeddings.embed_query("hello, world!")
 print(vector)
 ```
@@ -249,12 +253,9 @@ print("Answerable probability:", response.answerable_probability)
 
 ---
 
-
 ## Resources
 
-* [LangChain Documentation](https://docs.langchain.com/)
-* [Google Generative AI SDK](https://googleapis.github.io/python-genai/)
-* [Gemini Model Documentation](https://ai.google.dev/)
-
-
+- [LangChain Documentation](https://docs.langchain.com/)
+- [Google Generative AI SDK](https://googleapis.github.io/python-genai/)
+- [Gemini Model Documentation](https://ai.google.dev/)
 

{langchain_google_genai-2.1.7 → langchain_google_genai-2.1.9}/README.md

@@ -8,16 +8,20 @@ This package enables seamless access to Google Gemini's chat, vision, embeddings
 
 ## Table of Contents
 
-- [Overview](#overview)
-- [Installation](#installation)
-- [Quickstart](#quickstart)
-- [Chat Models](#chat-models)
-  - [Multimodal Inputs](#multimodal-inputs)
-  - [Multimodal Outputs](#multimodal-outputs)
-  - [Multimodal Outputs in Chains](#multimodal-outputs-in-chains)
-  - [Thinking Support](#thinking-support)
-- [Embeddings](#embeddings)
-- [Semantic Retrieval (RAG)](#semantic-retrieval-rag)
+- [langchain-google-genai](#langchain-google-genai)
+  - [Table of Contents](#table-of-contents)
+  - [Overview](#overview)
+  - [Installation](#installation)
+  - [Quickstart](#quickstart)
+  - [Chat Models](#chat-models)
+    - [Multimodal Inputs](#multimodal-inputs)
+    - [Multimodal Outputs](#multimodal-outputs)
+    - [Audio Output](#audio-output)
+    - [Multimodal Outputs in Chains](#multimodal-outputs-in-chains)
+    - [Thinking Support](#thinking-support)
+  - [Embeddings](#embeddings)
+  - [Semantic Retrieval (RAG)](#semantic-retrieval-rag)
+  - [Resources](#resources)
 
 ---
 
@@ -88,9 +92,9 @@ print(response.content)
 
 ✅ `image_url` can be:
 
-* A public image URL
-* A Google Cloud Storage path (`gcs://...`)
-* A base64-encoded image (e.g., `data:image/png;base64,...`)
+- A public image URL
+- A Google Cloud Storage path (`gcs://...`)
+- A base64-encoded image (e.g., `data:image/png;base64,...`)
 
 ---
 
@@ -187,7 +191,7 @@ You can use Gemini embeddings in LangChain:
 ```python
 from langchain_google_genai import GoogleGenerativeAIEmbeddings
 
-embeddings = GoogleGenerativeAIEmbeddings(model="models/embedding-001")
+embeddings = GoogleGenerativeAIEmbeddings(model="models/gemini-embedding-001")
 vector = embeddings.embed_query("hello, world!")
 print(vector)
 ```
@@ -228,11 +232,8 @@ print("Answerable probability:", response.answerable_probability)
 
 ---
 
-
 ## Resources
 
-* [LangChain Documentation](https://docs.langchain.com/)
-* [Google Generative AI SDK](https://googleapis.github.io/python-genai/)
-* [Gemini Model Documentation](https://ai.google.dev/)
-
-
+- [LangChain Documentation](https://docs.langchain.com/)
+- [Google Generative AI SDK](https://googleapis.github.io/python-genai/)
+- [Gemini Model Documentation](https://ai.google.dev/)

{langchain_google_genai-2.1.7 → langchain_google_genai-2.1.9}/langchain_google_genai/__init__.py

@@ -4,55 +4,57 @@ This module integrates Google's Generative AI models, specifically the Gemini se
 
 **Chat Models**
 
-The `ChatGoogleGenerativeAI` class is the primary interface for interacting with Google's Gemini chat models. It allows users to send and receive messages using a specified Gemini model, suitable for various conversational AI applications.
+The ``ChatGoogleGenerativeAI`` class is the primary interface for interacting with Google's Gemini chat models. It allows users to send and receive messages using a specified Gemini model, suitable for various conversational AI applications.
 
 **LLMs**
 
-The `GoogleGenerativeAI` class is the primary interface for interacting with Google's Gemini LLMs. It allows users to generate text using a specified Gemini model.
+The ``GoogleGenerativeAI`` class is the primary interface for interacting with Google's Gemini LLMs. It allows users to generate text using a specified Gemini model.
 
 **Embeddings**
 
-The `GoogleGenerativeAIEmbeddings` class provides functionalities to generate embeddings using Google's models.
+The ``GoogleGenerativeAIEmbeddings`` class provides functionalities to generate embeddings using Google's models.
 These embeddings can be used for a range of NLP tasks, including semantic analysis, similarity comparisons, and more.
+
 **Installation**
 
 To install the package, use pip:
 
-```python
-pip install -U langchain-google-genai
-```
-## Using Chat Models
+.. code-block:: python
+    pip install -U langchain-google-genai
+
+**Using Chat Models**
 
 After setting up your environment with the required API key, you can interact with the Google Gemini models.
 
-```python
-from langchain_google_genai import ChatGoogleGenerativeAI
+.. code-block:: python
+
+    from langchain_google_genai import ChatGoogleGenerativeAI
 
-llm = ChatGoogleGenerativeAI(model="gemini-pro")
-llm.invoke("Sing a ballad of LangChain.")
-```
+    llm = ChatGoogleGenerativeAI(model="gemini-2.5-pro")
+    llm.invoke("Sing a ballad of LangChain.")
 
-## Using LLMs
+**Using LLMs**
 
 The package also supports generating text with Google's models.
 
-```python
-from langchain_google_genai import GoogleGenerativeAI
+.. code-block:: python
 
-llm = GoogleGenerativeAI(model="gemini-pro")
-llm.invoke("Once upon a time, a library called LangChain")
-```
+    from langchain_google_genai import GoogleGenerativeAI
 
-## Embedding Generation
+    llm = GoogleGenerativeAI(model="gemini-2.5-pro")
+    llm.invoke("Once upon a time, a library called LangChain")
+
+**Embedding Generation**
 
 The package also supports creating embeddings with Google's models, useful for textual similarity and other NLP applications.
 
-```python
-from langchain_google_genai import GoogleGenerativeAIEmbeddings
+.. code-block:: python
+
+    from langchain_google_genai import GoogleGenerativeAIEmbeddings
+
+    embeddings = GoogleGenerativeAIEmbeddings(model="models/gemini-embedding-001")
+    embeddings.embed_query("hello, world!")
 
-embeddings = GoogleGenerativeAIEmbeddings(model="models/embedding-001")
-embeddings.embed_query("hello, world!")
-```
 """  # noqa: E501
 
 from langchain_google_genai._enums import HarmBlockThreshold, HarmCategory, Modality

{langchain_google_genai-2.1.7 → langchain_google_genai-2.1.9}/langchain_google_genai/_common.py

@@ -24,8 +24,8 @@ class _BaseGoogleGenerativeAI(BaseModel):
     model: str = Field(
         ...,
         description="""The name of the model to use.
-Supported examples:
-    - gemini-pro
+Examples:
+    - gemini-2.5-pro
     - models/text-bison-001""",
     )
     """Model name to use."""
@@ -39,20 +39,19 @@ Supported examples:
     "when making API calls. If not provided, credentials will be ascertained from "
     "the GOOGLE_API_KEY envvar"
     temperature: float = 0.7
-    """Run inference with this temperature. Must by in the closed interval
-       [0.0, 2.0]."""
+    """Run inference with this temperature. Must be within ``[0.0, 2.0]``."""
     top_p: Optional[float] = None
     """Decode using nucleus sampling: consider the smallest set of tokens whose
-       probability sum is at least top_p. Must be in the closed interval [0.0, 1.0]."""
+       probability sum is at least ``top_p``. Must be within ``[0.0, 1.0]``."""
     top_k: Optional[int] = None
-    """Decode using top-k sampling: consider the set of top_k most probable tokens.
+    """Decode using top-k sampling: consider the set of ``top_k`` most probable tokens.
        Must be positive."""
     max_output_tokens: Optional[int] = Field(default=None, alias="max_tokens")
     """Maximum number of tokens to include in a candidate. Must be greater than zero.
-       If unset, will default to 64."""
+       If unset, will default to ``64``."""
     n: int = 1
     """Number of chat completions to generate for each prompt. Note that the API may
-       not return the full n completions if duplicates are generated."""
+       not return the full ``n`` completions if duplicates are generated."""
     max_retries: int = 6
     """The maximum number of retries to make when generating."""
 
@@ -94,6 +93,7 @@ Supported examples:
     
         For example: 
 
+        .. code-block:: python
             from google.generativeai.types.safety_types import HarmBlockThreshold, HarmCategory
 
             safety_settings = {
@@ -102,7 +102,7 @@ Supported examples:
                 HarmCategory.HARM_CATEGORY_HARASSMENT: HarmBlockThreshold.BLOCK_LOW_AND_ABOVE,
                 HarmCategory.HARM_CATEGORY_SEXUALLY_EXPLICIT: HarmBlockThreshold.BLOCK_NONE,
             }
-            """  # noqa: E501
+    """  # noqa: E501
 
     @property
     def lc_secrets(self) -> Dict[str, str]:
@@ -149,7 +149,7 @@ def get_client_info(module: Optional[str] = None) -> "ClientInfo":
         module (Optional[str]):
             Optional. The module for a custom user agent header.
     Returns:
-        google.api_core.gapic_v1.client_info.ClientInfo
+        ``google.api_core.gapic_v1.client_info.ClientInfo``
     """
     client_library_version, user_agent = get_user_agent(module)
     return ClientInfo(

{langchain_google_genai-2.1.7 → langchain_google_genai-2.1.9}/langchain_google_genai/_function_utils.py

@@ -322,7 +322,7 @@ def _get_properties_from_schema_any(schema: Any) -> Dict[str, Any]:
 
 
 def _get_properties_from_schema(schema: Dict) -> Dict[str, Any]:
-    properties = {}
+    properties: Dict[str, Dict[str, Union[str, int, Dict, List]]] = {}
     for k, v in schema.items():
         if not isinstance(k, str):
             logger.warning(f"Key '{k}' is not supported in schema, type={type(k)}")
@@ -331,7 +331,14 @@ def _get_properties_from_schema(schema: Dict) -> Dict[str, Any]:
             logger.warning(f"Value '{v}' is not supported in schema, ignoring v={v}")
             continue
         properties_item: Dict[str, Union[str, int, Dict, List]] = {}
-        if v.get("type") or v.get("anyOf") or v.get("type_"):
+        if v.get("anyOf") and all(
+            anyOf_type.get("type") != "null" for anyOf_type in v.get("anyOf", [])
+        ):
+            properties_item["anyOf"] = [
+                _format_json_schema_to_gapic(anyOf_type)
+                for anyOf_type in v.get("anyOf", [])
+            ]
+        elif v.get("type") or v.get("anyOf") or v.get("type_"):
             item_type_ = _get_type_from_schema(v)
             properties_item["type_"] = item_type_
             if _is_nullable_schema(v):

{langchain_google_genai-2.1.7 → langchain_google_genai-2.1.9}/langchain_google_genai/_genai_extension.py

@@ -174,12 +174,12 @@ class TestCredentials(credentials.Credentials):
 
     @property
     def expired(self) -> bool:
-        """Returns `False`, test credentials never expire."""
+        """Returns ``False``, test credentials never expire."""
         return False
 
     @property
     def valid(self) -> bool:
-        """Returns `True`, test credentials are always valid."""
+        """Returns ``True``, test credentials are always valid."""
         return True
 
     def refresh(self, request: Any) -> None:
@@ -206,11 +206,11 @@ class TestCredentials(credentials.Credentials):
 def _get_credentials() -> Optional[credentials.Credentials]:
     """Returns credential from config if set or fake credentials for unit testing.
 
-    If _config.testing is True, a fake credential is returned.
+    If ``_config.testing`` is ``True``, a fake credential is returned.
     Otherwise, we are in a real environment and will use credentials if provided
-    or None is returned.
+    or ``None`` is returned.
 
-    If None is passed to the clients later on, the actual credentials will be
+    If ``None`` is passed to the clients later on, the actual credentials will be
     inferred by the rules specified in google.auth package.
     """
     if _config.testing:

{langchain_google_genai-2.1.7 → langchain_google_genai-2.1.9}/langchain_google_genai/_image_utils.py

@@ -30,7 +30,7 @@ class ImageBytesLoader:
     """
 
     def load_bytes(self, image_string: str) -> bytes:
-        """Routes to the correct loader based on the image_string.
+        """Routes to the correct loader based on the ``'image_string'``.
 
         Args:
             image_string: Can be either:
@@ -178,8 +178,8 @@ def image_bytes_to_b64_string(
 
     Args:
         image_bytes: Bytes of the image.
-        encoding: Type of encoding in the string. 'ascii' by default.
-        image_format: Format of the image. 'png' by default.
+        encoding: Type of encoding in the string. ``'ascii'`` by default.
+        image_format: Format of the image. ``'png'`` by default.
 
     Returns:
         B64 image encoded string.

{langchain_google_genai-2.1.7 → langchain_google_genai-2.1.9}/langchain_google_genai/chat_models.py

@@ -6,6 +6,7 @@ import io
 import json
 import logging
 import mimetypes
+import time
 import uuid
 import warnings
 import wave
@@ -31,7 +32,7 @@ from typing import (
 import filetype  # type: ignore[import]
 import google.api_core
 
-# TODO: remove ignore once the google package is published with types
+# TODO: remove ignore once the Google package is published with types
 import proto  # type: ignore[import]
 from google.ai.generativelanguage_v1beta import (
     GenerativeServiceAsyncClient as v1betaGenerativeServiceAsyncClient,
@@ -153,7 +154,12 @@ class ChatGoogleGenerativeAIError(GoogleGenerativeAIError):
     """
 
 
-def _create_retry_decorator() -> Callable[[Any], Any]:
+def _create_retry_decorator(
+    max_retries: int = 6,
+    wait_exponential_multiplier: float = 2.0,
+    wait_exponential_min: float = 1.0,
+    wait_exponential_max: float = 60.0,
+) -> Callable[[Any], Any]:
     """
     Creates and returns a preconfigured tenacity retry decorator.
 
@@ -165,15 +171,14 @@ def _create_retry_decorator() -> Callable[[Any], Any]:
         Callable[[Any], Any]: A retry decorator configured for handling specific
         Google API exceptions.
     """
-    multiplier = 2
-    min_seconds = 1
-    max_seconds = 60
-    max_retries = 2
-
     return retry(
         reraise=True,
         stop=stop_after_attempt(max_retries),
-        wait=wait_exponential(multiplier=multiplier, min=min_seconds, max=max_seconds),
+        wait=wait_exponential(
+            multiplier=wait_exponential_multiplier,
+            min=wait_exponential_min,
+            max=wait_exponential_max,
+        ),
         retry=(
             retry_if_exception_type(google.api_core.exceptions.ResourceExhausted)
             | retry_if_exception_type(google.api_core.exceptions.ServiceUnavailable)
@@ -198,13 +203,17 @@ def _chat_with_retry(generation_method: Callable, **kwargs: Any) -> Any:
     Returns:
         Any: The result from the chat generation method.
     """
-    retry_decorator = _create_retry_decorator()
+    retry_decorator = _create_retry_decorator(
+        max_retries=kwargs.get("max_retries", 6),
+        wait_exponential_multiplier=kwargs.get("wait_exponential_multiplier", 2.0),
+        wait_exponential_min=kwargs.get("wait_exponential_min", 1.0),
+        wait_exponential_max=kwargs.get("wait_exponential_max", 60.0),
+    )
 
     @retry_decorator
     def _chat_with_retry(**kwargs: Any) -> Any:
         try:
             return generation_method(**kwargs)
-        # Do not retry for these errors.
         except google.api_core.exceptions.FailedPrecondition as exc:
             if "location is not supported" in exc.message:
                 error_msg = (
@@ -218,6 +227,13 @@ def _chat_with_retry(generation_method: Callable, **kwargs: Any) -> Any:
             raise ChatGoogleGenerativeAIError(
                 f"Invalid argument provided to Gemini: {e}"
             ) from e
+        except google.api_core.exceptions.ResourceExhausted as e:
+            # Handle quota-exceeded error with recommended retry delay
+            if hasattr(e, "retry_after") and e.retry_after < kwargs.get(
+                "wait_exponential_max", 60.0
+            ):
+                time.sleep(e.retry_after)
+            raise e
         except Exception as e:
             raise e
 
@@ -295,7 +311,7 @@ def _is_openai_image_block(block: dict) -> bool:
 def _convert_to_parts(
     raw_content: Union[str, Sequence[Union[str, dict]]],
 ) -> List[Part]:
-    """Converts a list of LangChain messages into a google parts."""
+    """Converts a list of LangChain messages into a Google parts."""
     parts = []
     content = [raw_content] if isinstance(raw_content, str) else raw_content
     image_loader = ImageBytesLoader()
@@ -413,7 +429,7 @@ def _convert_to_parts(
 def _convert_tool_message_to_parts(
     message: ToolMessage | FunctionMessage, name: Optional[str] = None
 ) -> list[Part]:
-    """Converts a tool or function message to a google part."""
+    """Converts a tool or function message to a Google part."""
     # Legacy agent stores tool name in message.additional_kwargs instead of message.name
     name = message.name or name or message.additional_kwargs.get("name")
     response: Any
@@ -824,14 +840,13 @@ class ChatGoogleGenerativeAI(_BaseGoogleGenerativeAI, BaseChatModel):
         To use, you must have either:
 
             1. The ``GOOGLE_API_KEY`` environment variable set with your API key, or
-            2. Pass your API key using the google_api_key kwarg
-            to the ChatGoogleGenerativeAI constructor.
+            2. Pass your API key using the ``google_api_key`` kwarg to the ChatGoogleGenerativeAI constructor.
 
         .. code-block:: python
 
             from langchain_google_genai import ChatGoogleGenerativeAI
 
-            llm = ChatGoogleGenerativeAI(model="gemini-2.0-flash-001")
+            llm = ChatGoogleGenerativeAI(model="gemini-2.5-flash")
             llm.invoke("Write me a ballad about LangChain")
 
     Invoke:
@@ -893,8 +908,8 @@ class ChatGoogleGenerativeAI(_BaseGoogleGenerativeAI, BaseChatModel):
 
     Context Caching:
         Context caching allows you to store and reuse content (e.g., PDFs, images) for faster processing.
-        The `cached_content` parameter accepts a cache name created via the Google Generative AI API.
-        Below are two examples: caching a single file directly and caching multiple files using `Part`.
+        The ``cached_content`` parameter accepts a cache name created via the Google Generative AI API.
+        Below are two examples: caching a single file directly and caching multiple files using ``Part``.
 
         Single File Example:
         This caches a single file and queries it.
@@ -1041,7 +1056,7 @@ class ChatGoogleGenerativeAI(_BaseGoogleGenerativeAI, BaseChatModel):
         .. code-block:: python
 
             from google.ai.generativelanguage_v1beta.types import Tool as GenAITool
-            llm = ChatGoogleGenerativeAI(model="gemini-2.0-flash-exp")
+            llm = ChatGoogleGenerativeAI(model="gemini-2.5-flash")
             resp = llm.invoke(
                 "When is the next total solar eclipse in US?",
                 tools=[GenAITool(google_search={})],
@@ -1099,6 +1114,144 @@ class ChatGoogleGenerativeAI(_BaseGoogleGenerativeAI, BaseChatModel):
 
             'The weather in this image appears to be sunny and pleasant. The sky is a bright blue with scattered white clouds, suggesting fair weather. The lush green grass and trees indicate a warm and possibly slightly breezy day. There are no signs of rain or storms.'
 
+    PDF input:
+        .. code-block:: python
+
+            import base64
+            from langchain_core.messages import HumanMessage
+
+            pdf_bytes = open("/path/to/your/test.pdf", 'rb').read()
+            pdf_base64 = base64.b64encode(pdf_bytes).decode('utf-8')
+
+            message = HumanMessage(
+                content=[
+                    {"type": "text", "text": "describe the document in a sentence"},
+                    {
+                        "type": "file",
+                        "source_type": "base64",
+                        "mime_type":"application/pdf",
+                        "data": pdf_base64
+                    }
+                ]
+            )
+            ai_msg = llm.invoke([message])
+            ai_msg.content
+
+        .. code-block:: python
+
+            'This research paper describes a system developed for SemEval-2025 Task 9, which aims to automate the detection of food hazards from recall reports, addressing the class imbalance problem by leveraging LLM-based data augmentation techniques and transformer-based models to improve performance.'
+
+    Video input:
+        .. code-block:: python
+
+            import base64
+            from langchain_core.messages import HumanMessage
+
+            video_bytes = open("/path/to/your/video.mp4", 'rb').read()
+            video_base64 = base64.b64encode(video_bytes).decode('utf-8')
+
+            message = HumanMessage(
+                content=[
+                    {"type": "text", "text": "describe what's in this video in a sentence"},
+                    {
+                        "type": "file",
+                        "source_type": "base64",
+                        "mime_type": "video/mp4",
+                        "data": video_base64
+                    }
+                ]
+            )
+            ai_msg = llm.invoke([message])
+            ai_msg.content
+
+        .. code-block:: python
+
+            'Tom and Jerry, along with a turkey, engage in a chaotic Thanksgiving-themed adventure involving a corn-on-the-cob chase, maze antics, and a disastrous attempt to prepare a turkey dinner.'
+
+        You can also pass YouTube URLs directly:
+
+        .. code-block:: python
+
+            from langchain_core.messages import HumanMessage
+
+            message = HumanMessage(
+                content=[
+                    {"type": "text", "text": "summarize the video in 3 sentences."},
+                    {
+                        "type": "media",
+                        "file_uri": "https://www.youtube.com/watch?v=9hE5-98ZeCg",
+                        "mime_type": "video/mp4",
+                    }
+                ]
+            )
+            ai_msg = llm.invoke([message])
+            ai_msg.content
+
+        .. code-block:: python
+
+            'The video is a demo of multimodal live streaming in Gemini 2.0. The narrator is sharing his screen in AI Studio and asks if the AI can see it. The AI then reads text that is highlighted on the screen, defines the word “multimodal,” and summarizes everything that was seen and heard.'
+
+    Audio input:
+        .. code-block:: python
+
+            import base64
+            from langchain_core.messages import HumanMessage
+
+            audio_bytes = open("/path/to/your/audio.mp3", 'rb').read()
+            audio_base64 = base64.b64encode(audio_bytes).decode('utf-8')
+
+            message = HumanMessage(
+                content=[
+                    {"type": "text", "text": "summarize this audio in a sentence"},
+                    {
+                        "type": "file",
+                        "source_type": "base64",
+                        "mime_type":"audio/mp3",
+                        "data": audio_base64
+                    }
+                ]
+            )
+            ai_msg = llm.invoke([message])
+            ai_msg.content
+
+        .. code-block:: python
+
+            "In this episode of the Made by Google podcast, Stephen Johnson and Simon Tokumine discuss NotebookLM, a tool designed to help users understand complex material in various modalities, with a focus on its unexpected uses, the development of audio overviews, and the implementation of new features like mind maps and source discovery."
+
+    File upload (URI-based):
+        You can also upload files to Google's servers and reference them by URI.
+        This works for PDFs, images, videos, and audio files.
+
+        .. code-block:: python
+
+            import time
+            from google import genai
+            from langchain_core.messages import HumanMessage
+
+            client = genai.Client()
+
+            myfile = client.files.upload(file="/path/to/your/sample.pdf")
+            while myfile.state.name == "PROCESSING":
+                time.sleep(2)
+                myfile = client.files.get(name=myfile.name)
+
+            message = HumanMessage(
+                content=[
+                    {"type": "text", "text": "What is in the document?"},
+                    {
+                        "type": "media",
+                        "file_uri": myfile.uri,
+                        "mime_type": "application/pdf",
+                    },
+                ]
+            )
+            ai_msg = llm.invoke([message])
+            ai_msg.content
+
+        .. code-block:: python
+
+            "This research paper assesses and mitigates multi-turn jailbreak vulnerabilities in large language models using the Crescendo attack study, evaluating attack success rates and mitigation strategies like prompt hardening and LLM-as-guardrail."
+
     Token usage:
         .. code-block:: python
 
@@ -1140,12 +1293,15 @@ class ChatGoogleGenerativeAI(_BaseGoogleGenerativeAI, BaseChatModel):
 
     response_mime_type: Optional[str] = None
     """Optional. Output response mimetype of the generated candidate text. Only
-        supported in Gemini 1.5 and later models. Supported mimetype:
-            * "text/plain": (default) Text output.
-            * "application/json": JSON response in the candidates.
-            * "text/x.enum": Enum in plain text.
-       The model also needs to be prompted to output the appropriate response
-       type, otherwise the behavior is undefined. This is a preview feature.
+    supported in Gemini 1.5 and later models.
+    
+    Supported mimetype:
+        * ``'text/plain'``: (default) Text output.
+        * ``'application/json'``: JSON response in the candidates.
+        * ``'text/x.enum'``: Enum in plain text.
+    
+    The model also needs to be prompted to output the appropriate response
+    type, otherwise the behavior is undefined. This is a preview feature.
     """
 
     response_schema: Optional[Dict[str, Any]] = None
@@ -1230,9 +1386,7 @@ class ChatGoogleGenerativeAI(_BaseGoogleGenerativeAI, BaseChatModel):
         if self.top_k is not None and self.top_k <= 0:
             raise ValueError("top_k must be positive")
 
-        if not any(
-            self.model.startswith(prefix) for prefix in ("models/", "tunedModels/")
-        ):
+        if not any(self.model.startswith(prefix) for prefix in ("models/",)):
             self.model = f"models/{self.model}"
 
         additional_headers = self.additional_headers or {}
@@ -1328,7 +1482,7 @@ class ChatGoogleGenerativeAI(_BaseGoogleGenerativeAI, BaseChatModel):
 
             else:
                 raise ValueError(
-                    "Tools are already defined." "code_execution tool can't be defined"
+                    "Tools are already defined.code_execution tool can't be defined"
                 )
 
         return super().invoke(input, config, stop=stop, **kwargs)
@@ -1545,7 +1699,7 @@ class ChatGoogleGenerativeAI(_BaseGoogleGenerativeAI, BaseChatModel):
             )
 
             if run_manager:
-                run_manager.on_llm_new_token(gen.text)
+                run_manager.on_llm_new_token(gen.text, chunk=gen)
             yield gen
 
     async def _astream(
@@ -1611,7 +1765,7 @@ class ChatGoogleGenerativeAI(_BaseGoogleGenerativeAI, BaseChatModel):
                 )
 
                 if run_manager:
-                    await run_manager.on_llm_new_token(gen.text)
+                    await run_manager.on_llm_new_token(gen.text, chunk=gen)
                 yield gen
 
     def _prepare_request(
@@ -1816,7 +1970,8 @@ class ChatGoogleGenerativeAI(_BaseGoogleGenerativeAI, BaseChatModel):
             tools: A list of tool definitions to bind to this chat model.
                 Can be a pydantic model, callable, or BaseTool. Pydantic
                 models, callables, and BaseTools will be automatically converted to
-                their schema dictionary representation.
+                their schema dictionary representation. Tools with Union types in
+                their arguments are now supported and converted to `anyOf` schemas.
             **kwargs: Any additional parameters to pass to the
                 :class:`~langchain.runnable.Runnable` constructor.
         """

{langchain_google_genai-2.1.7 → langchain_google_genai-2.1.9}/langchain_google_genai/embeddings.py

@@ -17,7 +17,10 @@ from langchain_google_genai._common import (
     GoogleGenerativeAIError,
     get_client_info,
 )
-from langchain_google_genai._genai_extension import build_generative_service
+from langchain_google_genai._genai_extension import (
+    build_generative_async_service,
+    build_generative_service,
+)
 
 _MAX_TOKENS_PER_BATCH = 20000
 _DEFAULT_BATCH_SIZE = 100
@@ -29,29 +32,30 @@ class GoogleGenerativeAIEmbeddings(BaseModel, Embeddings):
     To use, you must have either:
 
         1. The ``GOOGLE_API_KEY`` environment variable set with your API key, or
-        2. Pass your API key using the google_api_key kwarg
-        to the GoogleGenerativeAIEmbeddings constructor.
+        2. Pass your API key using the google_api_key kwarg to the
+        GoogleGenerativeAIEmbeddings constructor.
 
     Example:
         .. code-block:: python
 
             from langchain_google_genai import GoogleGenerativeAIEmbeddings
 
-            embeddings = GoogleGenerativeAIEmbeddings(model="models/embedding-001")
+            embeddings = GoogleGenerativeAIEmbeddings(model="gemini-embedding-001")
             embeddings.embed_query("What's our Q1 revenue?")
     """
 
     client: Any = None  #: :meta private:
+    async_client: Any = None  #: :meta private:
     model: str = Field(
         ...,
         description="The name of the embedding model to use. "
-        "Example: models/embedding-001",
+        "Example: ``'models/gemini-embedding-001'``",
     )
     task_type: Optional[str] = Field(
         default=None,
         description="The task type. Valid options include: "
-        "task_type_unspecified, retrieval_query, retrieval_document, "
-        "semantic_similarity, classification, and clustering",
+        "``'task_type_unspecified'``, ``'retrieval_query'``, ``'retrieval_document'``, "
+        "``'semantic_similarity'``, ``'classification'``, and ``'clustering'``",
     )
     google_api_key: Optional[SecretStr] = Field(
         default_factory=secret_from_env("GOOGLE_API_KEY", default=None),
@@ -76,7 +80,7 @@ class GoogleGenerativeAIEmbeddings(BaseModel, Embeddings):
     )
     transport: Optional[str] = Field(
         default=None,
-        description="A string, one of: [`rest`, `grpc`, `grpc_asyncio`].",
+        description="A string, one of: [``'rest'``, ``'grpc'``, ``'grpc_asyncio'``].",
     )
     request_options: Optional[Dict] = Field(
         default=None,
@@ -93,6 +97,9 @@ class GoogleGenerativeAIEmbeddings(BaseModel, Embeddings):
             google_api_key = self.google_api_key
         client_info = get_client_info("GoogleGenerativeAIEmbeddings")
 
+        if not any(self.model.startswith(prefix) for prefix in ("models/",)):
+            self.model = f"models/{self.model}"
+
         self.client = build_generative_service(
             credentials=self.credentials,
             api_key=google_api_key,
@@ -100,6 +107,13 @@ class GoogleGenerativeAIEmbeddings(BaseModel, Embeddings):
             client_options=self.client_options,
             transport=self.transport,
         )
+        self.async_client = build_generative_async_service(
+            credentials=self.credentials,
+            api_key=google_api_key,
+            client_info=client_info,
+            client_options=self.client_options,
+            transport=self.transport,
+        )
         return self
 
     @staticmethod
@@ -166,12 +180,12 @@ class GoogleGenerativeAIEmbeddings(BaseModel, Embeddings):
     def _prepare_request(
         self,
         text: str,
+        *,
         task_type: Optional[str] = None,
         title: Optional[str] = None,
         output_dimensionality: Optional[int] = None,
     ) -> EmbedContentRequest:
         task_type = self.task_type or task_type or "RETRIEVAL_DOCUMENT"
-        # https://ai.google.dev/api/rest/v1/models/batchEmbedContents#EmbedContentRequest
         request = EmbedContentRequest(
             content={"parts": [{"text": text}]},
             model=self.model,
@@ -190,17 +204,17 @@ class GoogleGenerativeAIEmbeddings(BaseModel, Embeddings):
         titles: Optional[List[str]] = None,
         output_dimensionality: Optional[int] = None,
     ) -> List[List[float]]:
-        """Embed a list of strings. Google Generative AI currently
-        sets a max batch size of 100 strings.
+        """Embed a list of strings using the `batch endpoint <https://ai.google.dev/api/embeddings#method:-models.batchembedcontents>`__.
+
+        Google Generative AI currently sets a max batch size of 100 strings.
 
         Args:
             texts: List[str] The list of strings to embed.
             batch_size: [int] The batch size of embeddings to send to the model
-            task_type: task_type (https://ai.google.dev/api/rest/v1/TaskType)
+            task_type: `task_type <https://ai.google.dev/api/embeddings#tasktype>`__
             titles: An optional list of titles for texts provided.
-            Only applicable when TaskType is RETRIEVAL_DOCUMENT.
-            output_dimensionality: Optional reduced dimension for the output embedding.
-            https://ai.google.dev/api/rest/v1/models/batchEmbedContents#EmbedContentRequest
+              Only applicable when TaskType is ``'RETRIEVAL_DOCUMENT'``.
+            output_dimensionality: Optional `reduced dimension for the output embedding <https://ai.google.dev/api/embeddings#EmbedContentRequest>`__.
         Returns:
             List of embeddings, one for each text.
         """
@@ -237,26 +251,26 @@ class GoogleGenerativeAIEmbeddings(BaseModel, Embeddings):
     def embed_query(
         self,
         text: str,
+        *,
         task_type: Optional[str] = None,
         title: Optional[str] = None,
         output_dimensionality: Optional[int] = None,
     ) -> List[float]:
-        """Embed a text, using the non-batch endpoint:
-        https://ai.google.dev/api/rest/v1/models/embedContent#EmbedContentRequest
+        """Embed a text, using the `non-batch endpoint <https://ai.google.dev/api/embeddings#method:-models.embedcontent>`__.
 
         Args:
             text: The text to embed.
-            task_type: task_type (https://ai.google.dev/api/rest/v1/TaskType)
+            task_type: `task_type <https://ai.google.dev/api/embeddings#tasktype>`__
             title: An optional title for the text.
-            Only applicable when TaskType is RETRIEVAL_DOCUMENT.
-            output_dimensionality: Optional reduced dimension for the output embedding.
+              Only applicable when TaskType is ``'RETRIEVAL_DOCUMENT'``.
+            output_dimensionality: Optional `reduced dimension for the output embedding <https://ai.google.dev/api/embeddings#EmbedContentRequest>`__.
 
         Returns:
             Embedding for the text.
         """
         task_type_to_use = task_type if task_type else self.task_type
         if task_type_to_use is None:
-            task_type_to_use = "RETRIEVAL_QUERY"  # Default to RETRIEVAL_QUERY
+            task_type_to_use = "RETRIEVAL_QUERY"
         try:
             request: EmbedContentRequest = self._prepare_request(
                 text=text,
@@ -268,3 +282,93 @@ class GoogleGenerativeAIEmbeddings(BaseModel, Embeddings):
         except Exception as e:
             raise GoogleGenerativeAIError(f"Error embedding content: {e}") from e
         return list(result.embedding.values)
+
+    async def aembed_documents(
+        self,
+        texts: List[str],
+        *,
+        batch_size: int = _DEFAULT_BATCH_SIZE,
+        task_type: Optional[str] = None,
+        titles: Optional[List[str]] = None,
+        output_dimensionality: Optional[int] = None,
+    ) -> List[List[float]]:
+        """Embed a list of strings using the `batch endpoint <https://ai.google.dev/api/embeddings#method:-models.batchembedcontents>`__.
+
+        Google Generative AI currently sets a max batch size of 100 strings.
+
+        Args:
+            texts: List[str] The list of strings to embed.
+            batch_size: [int] The batch size of embeddings to send to the model
+            task_type: `task_type <https://ai.google.dev/api/embeddings#tasktype>`__
+            titles: An optional list of titles for texts provided.
+                Only applicable when TaskType is ``'RETRIEVAL_DOCUMENT'``.
+            output_dimensionality: Optional `reduced dimension for the output embedding <https://ai.google.dev/api/embeddings#EmbedContentRequest>`__.
+        Returns:
+            List of embeddings, one for each text.
+        """
+        embeddings: List[List[float]] = []
+        batch_start_index = 0
+        for batch in GoogleGenerativeAIEmbeddings._prepare_batches(texts, batch_size):
+            if titles:
+                titles_batch = titles[
+                    batch_start_index : batch_start_index + len(batch)
+                ]
+                batch_start_index += len(batch)
+            else:
+                titles_batch = [None] * len(batch)  # type: ignore[list-item]
+
+            requests = [
+                self._prepare_request(
+                    text=text,
+                    task_type=task_type,
+                    title=title,
+                    output_dimensionality=output_dimensionality,
+                )
+                for text, title in zip(batch, titles_batch)
+            ]
+
+            try:
+                result = await self.async_client.batch_embed_contents(
+                    BatchEmbedContentsRequest(requests=requests, model=self.model)
+                )
+            except Exception as e:
+                raise GoogleGenerativeAIError(f"Error embedding content: {e}") from e
+            embeddings.extend([list(e.values) for e in result.embeddings])
+        return embeddings
+
+    async def aembed_query(
+        self,
+        text: str,
+        *,
+        task_type: Optional[str] = None,
+        title: Optional[str] = None,
+        output_dimensionality: Optional[int] = None,
+    ) -> List[float]:
+        """Embed a text, using the `non-batch endpoint <https://ai.google.dev/api/embeddings#method:-models.embedcontent>`__.
+
+        Args:
+            text: The text to embed.
+            task_type: `task_type <https://ai.google.dev/api/embeddings#tasktype>`__
+            title: An optional title for the text.
+                Only applicable when TaskType is ``'RETRIEVAL_DOCUMENT'``.
+            output_dimensionality: Optional `reduced dimension for the output embedding <https://ai.google.dev/api/embeddings#EmbedContentRequest>`__.
+
+        Returns:
+            Embedding for the text.
+        """
+        task_type_to_use = task_type if task_type else self.task_type
+        if task_type_to_use is None:
+            task_type_to_use = "RETRIEVAL_QUERY"
+        try:
+            request: EmbedContentRequest = self._prepare_request(
+                text=text,
+                task_type=task_type,
+                title=title,
+                output_dimensionality=output_dimensionality,
+            )
+            result: EmbedContentResponse = await self.async_client.embed_content(
+                request
+            )
+        except Exception as e:
+            raise GoogleGenerativeAIError(f"Error embedding content: {e}") from e
+        return list(result.embedding.values)

{langchain_google_genai-2.1.7 → langchain_google_genai-2.1.9}/langchain_google_genai/llms.py

@@ -29,7 +29,7 @@ class GoogleGenerativeAI(_BaseGoogleGenerativeAI, BaseLLM):
         .. code-block:: python
 
             from langchain_google_genai import GoogleGenerativeAI
-            llm = GoogleGenerativeAI(model="gemini-pro")
+            llm = GoogleGenerativeAI(model="gemini-2.5-pro")
     """
 
     client: Any = None  #: :meta private:
@@ -63,6 +63,9 @@ class GoogleGenerativeAI(_BaseGoogleGenerativeAI, BaseLLM):
     def validate_environment(self) -> Self:
         """Validates params and passes them to google-generativeai package."""
 
+        if not any(self.model.startswith(prefix) for prefix in ("models/",)):
+            self.model = f"models/{self.model}"
+
         self.client = ChatGoogleGenerativeAI(
             api_key=self.google_api_key,
             credentials=self.credentials,
@@ -86,6 +89,15 @@ class GoogleGenerativeAI(_BaseGoogleGenerativeAI, BaseLLM):
         """Get standard params for tracing."""
         ls_params = super()._get_ls_params(stop=stop, **kwargs)
         ls_params["ls_provider"] = "google_genai"
+
+        models_prefix = "models/"
+        ls_model_name = (
+            self.model[len(models_prefix) :]
+            if self.model and self.model.startswith(models_prefix)
+            else self.model
+        )
+        ls_params["ls_model_name"] = ls_model_name
+
         if ls_max_tokens := kwargs.get("max_output_tokens", self.max_output_tokens):
             ls_params["ls_max_tokens"] = ls_max_tokens
         return ls_params

{langchain_google_genai-2.1.7 → langchain_google_genai-2.1.9}/pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "langchain-google-genai"
-version = "2.1.7"
+version = "2.1.9"
 description = "An integration package connecting Google's genai package and LangChain"
 authors = []
 readme = "README.md"