通过 Gemini Developer API 或 Vertex AI 访问 Google 的生成式 AI 模型，包括 Gemini 系列。Gemini Developer API 提供快速设置，使用 API 密钥，非常适合个人开发者。Vertex AI 提供企业级功能，并与 Google Cloud Platform 集成。有关最新模型、模型 ID、其功能、上下文窗口等信息，请访问 Google AI 文档。
Vertex AI 整合与兼容性自 langchain-google-genai 4.0.0 起，此包使用整合的 google-genai SDK，取代了旧的 google-ai-generativelanguage SDK。此迁移通过 Gemini Developer API 和 Vertex AI 中的 Gemini API 提供对 Gemini 模型的支持，取代了 langchain-google-vertexai 中的某些类，例如 ChatVertexAI。请阅读完整公告和迁移指南。
messages = [ ( "system", "You are a helpful assistant that translates English to French. Translate the user sentence.", ), ("human", "I love programming."),]ai_msg = model.invoke(messages)ai_msg
from langchain_google_genai import ChatGoogleGenerativeAI# 在实例化时设置（应用于所有调用）model = ChatGoogleGenerativeAI( model="gemini-2.5-flash-image", image_config={"aspect_ratio": "16:9"},)# 或每次调用覆盖response = model.invoke( "Generate a photorealistic image of a cuddly cat wearing a hat.", image_config={"aspect_ratio": "1:1"},)
默认情况下，图像生成模型可能同时返回文本和图像（例如 “Ok! Here’s an image of a…”）。您可以通过设置 response_modalities 参数来请求模型仅返回图像：
from langchain_google_genai import ChatGoogleGenerativeAI, Modalitymodel = ChatGoogleGenerativeAI( model="gemini-2.5-flash-image", response_modalities=[Modality.IMAGE],)# 所有调用将仅返回图像response = model.invoke("Generate a photorealistic image of a cuddly cat wearing a hat.")
Vertex AI 限制音频生成模型目前在 Vertex AI 上处于有限预览状态，可能需要允许列表访问。如果在使用 TTS 模型时遇到 INVALID_ARGUMENT 错误且 vertexai=True，您的 GCP 项目可能需要被列入允许列表。有关更多详细信息，请参阅此 Google AI 论坛讨论。
from langchain_google_genai import ChatGoogleGenerativeAImodel = ChatGoogleGenerativeAI(model="gemini-2.5-flash-preview-tts")response = model.invoke("Please say The quick brown fox jumps over the lazy dog")# 音频的 Base64 编码二进制数据wav_data = response.additional_kwargs.get("audio")with open("output.wav", "wb") as f: f.write(wav_data)
from langchain_google_genai import ChatGoogleGenerativeAIfrom pydantic import BaseModelfrom typing import Literalclass Feedback(BaseModel): sentiment: Literal["positive", "neutral", "negative"] summary: strmodel = ChatGoogleGenerativeAI(model="gemini-3.1-pro-preview")structured_model = model.with_structured_output( schema=Feedback.model_json_schema(), method="json_schema")response = structured_model.invoke("The new UI is great!")response["sentiment"] # "positive"response["summary"] # "The user expresses positive..."
对于流式结构化输出，请合并字典而不是使用 +=：
stream = structured_model.stream("The interface is intuitive and beautiful!")full = next(stream)for chunk in stream: full.update(chunk) # 合并字典print(full) # 完整的结构化响应# -> {'sentiment': 'positive', 'summary': 'The user praises...'}
from langchain_google_genai import ChatGoogleGenerativeAImodel = ChatGoogleGenerativeAI(model="gemini-3.1-pro-preview")result = model.invoke("Explain the concept of prompt engineering in one sentence.")print(result.content)print("\nUsage Metadata:")print(result.usage_metadata)
Prompt engineering is the art and science of crafting effective text prompts to elicit desired and accurate responses from large language models.Usage Metadata:{'input_tokens': 10, 'output_tokens': 24, 'total_tokens': 34, 'input_token_details': {'cache_read': 0}}
from langchain_google_genai import ChatGoogleGenerativeAI# Gemini 3+: 使用 thinking_levelllm = ChatGoogleGenerativeAI( model="gemini-3.1-pro-preview", thinking_level="low",)response = llm.invoke("How many O's are in Google?")
from langchain_google_genai import ChatGoogleGenerativeAIllm = ChatGoogleGenerativeAI( model="gemini-3.1-pro-preview", include_thoughts=True,)response = llm.invoke("How many O's are in Google? How did you verify your answer?")reasoning_tokens = response.usage_metadata["output_token_details"]["reasoning"]print("Response:", response.content)print("Reasoning tokens used:", reasoning_tokens)
from langchain_google_genai import ChatGoogleGenerativeAImodel = ChatGoogleGenerativeAI(model="gemini-3.1-pro-preview")model_with_search = model.bind_tools([{"google_search": {}}])response = model_with_search.invoke("When is the next total solar eclipse in US?")response.content_blocks
[{'type': 'text', 'text': 'The next total solar eclipse visible in the contiguous United States will occur on...', 'annotations': [{'type': 'citation', 'id': 'abc123', 'url': '<url for source 1>', 'title': '<source 1 title>', 'start_index': 0, 'end_index': 99, 'cited_text': 'The next total solar eclipse...', 'extras': {'google_ai_metadata': {'web_search_queries': ['next total solar eclipse in US'], 'grounding_chunk_index': 0, 'confidence_scores': []}}}, ...
某些模型支持使用 Google 地图进行基础。地图基础将 Gemini 的生成功能与 Google 地图当前、事实的位置数据连接起来。这使得位置感知应用程序能够提供准确、特定于地理位置的响应。有关详细信息，请参阅 Gemini 文档。
from langchain_google_genai import ChatGoogleGenerativeAImodel = ChatGoogleGenerativeAI(model="gemini-2.5-pro")model_with_maps = model.bind_tools([{"google_maps": {}}])response = model_with_maps.invoke( "What are some good Italian restaurants near the Eiffel Tower in Paris?")
响应将包含来自 Google 地图的位置信息的基础元数据。您可以选择使用 tool_config 和 lat_lng 提供特定的位置上下文。当您希望相对于特定地理点进行基础查询时，这很有用。
from langchain_google_genai import ChatGoogleGenerativeAImodel = ChatGoogleGenerativeAI(model="gemini-2.5-pro")# 提供位置上下文（纬度和经度）model_with_maps = model.bind_tools( [{"google_maps": {}}], tool_config={ "retrieval_config": { # 埃菲尔铁塔 "lat_lng": { "latitude": 48.858844, "longitude": 2.294351, } } },)response = model_with_maps.invoke( "What Italian restaurants are within a 5 minute walk from here?")
上下文缓存允许您存储和重用内容（例如 PDF、图像）以加快处理速度。cached_content 参数接受通过 Google Generative AI API 创建的缓存名称。
单个文件示例
此示例缓存单个文件并对其进行查询。
import timefrom google import genaifrom google.genai import typesfrom langchain.messages import HumanMessagefrom langchain_google_genai import ChatGoogleGenerativeAIclient = genai.Client()# 上传文件file = client.files.upload(file="path/to/your/file")while file.state.name == "PROCESSING": time.sleep(2) file = client.files.get(name=file.name)# 创建缓存model = "gemini-3.1-pro-preview"cache = client.caches.create( model=model, config=types.CreateCachedContentConfig( display_name="Cached Content", system_instruction=( "You are an expert content analyzer, and your job is to answer " "the user's query based on the file you have access to." ), contents=[file], ttl="300s", ),)# 使用 LangChain 查询llm = ChatGoogleGenerativeAI( model=model, cached_content=cache.name,)message = HumanMessage(content="Summarize the main points of the content.")llm.invoke([message])
多个文件示例
此示例使用 Part 缓存两个文件并一起查询它们。
import timefrom google import genaifrom google.genai.types import CreateCachedContentConfig, Content, Partfrom langchain.messages import HumanMessagefrom langchain_google_genai import ChatGoogleGenerativeAIclient = genai.Client()# 上传文件file_1 = client.files.upload(file="./file1")while file_1.state.name == "PROCESSING": time.sleep(2) file_1 = client.files.get(name=file_1.name)file_2 = client.files.upload(file="./file2")while file_2.state.name == "PROCESSING": time.sleep(2) file_2 = client.files.get(name=file_2.name)# 使用多个文件创建缓存contents = [ Content( role="user", parts=[ Part.from_uri(file_uri=file_1.uri, mime_type=file_1.mime_type), Part.from_uri(file_uri=file_2.uri, mime_type=file_2.mime_type), ], )]model = "gemini-3.1-pro-preview"cache = client.caches.create( model=model, config=CreateCachedContentConfig( display_name="Cached Contents", system_instruction=( "You are an expert content analyzer, and your job is to answer " "the user's query based on the files you have access to." ), contents=contents, ttl="300s", ),)# 使用 LangChain 查询llm = ChatGoogleGenerativeAI( model=model, cached_content=cache.name,)message = HumanMessage( content="Provide a summary of the key information across both files.")llm.invoke([message])
