程序包 io.agentscope.core.model

类 GeminiChatModel

java.lang.Object

io.agentscope.core.model.ChatModelBase

io.agentscope.core.model.GeminiChatModel

所有已实现的接口:

Model

public class GeminiChatModel
extends ChatModelBase

Gemini Chat Model implementation using the official Google GenAI Java SDK.

This implementation provides complete integration with Gemini's Content Generation API, including tool calling and multi-agent conversation support.

Supported Features:

• Text generation with streaming and non-streaming modes
• Tool/function calling support
• Multi-agent conversation with history merging
• Vision capabilities (images, audio, video)
• Thinking mode (extended reasoning)

嵌套类概要

嵌套类

修饰符和类型	类	说明
static class	GeminiChatModel.Builder	Builder for creating GeminiChatModel instances.

构造器概要

构造器

构造器	说明
GeminiChatModel(String apiKey, String modelName, boolean streamEnabled, String project, String location, Boolean vertexAI, com.google.genai.types.HttpOptions httpOptions, com.google.auth.oauth2.GoogleCredentials credentials, com.google.genai.types.ClientOptions clientOptions, GenerateOptions defaultOptions, Formatter<com.google.genai.types.Content,com.google.genai.types.GenerateContentResponse,com.google.genai.types.GenerateContentConfig.Builder> formatter)	Creates a new Gemini chat model instance using the default endpoint configuration.
GeminiChatModel(String apiKey, String baseUrl, String modelName, boolean streamEnabled, String project, String location, Boolean vertexAI, com.google.genai.types.HttpOptions httpOptions, com.google.auth.oauth2.GoogleCredentials credentials, com.google.genai.types.ClientOptions clientOptions, GenerateOptions defaultOptions, Formatter<com.google.genai.types.Content,com.google.genai.types.GenerateContentResponse,com.google.genai.types.GenerateContentConfig.Builder> formatter)	Creates a new Gemini chat model instance.

方法概要

修饰符和类型	方法	说明
static GeminiChatModel.Builder	builder()	Creates a new builder for GeminiChatModel.
void	close()	Close the Gemini client.
protected reactor.core.publisher.Flux<ChatResponse>	doStream(List<Msg> messages, List<ToolSchema> tools, GenerateOptions options)	Stream chat completion responses from Gemini's API.
String	getModelName()	Get model name for logging and identification.

从类继承的方法 io.agentscope.core.model.ChatModelBase

stream

从类继承的方法 java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

构造器详细资料

GeminiChatModel

public GeminiChatModel(String apiKey,
 String baseUrl,
 String modelName,
 boolean streamEnabled,
 String project,
 String location,
 Boolean vertexAI,
 com.google.genai.types.HttpOptions httpOptions,
 com.google.auth.oauth2.GoogleCredentials credentials,
 com.google.genai.types.ClientOptions clientOptions,
 GenerateOptions defaultOptions,
 Formatter<com.google.genai.types.Content,com.google.genai.types.GenerateContentResponse,com.google.genai.types.GenerateContentConfig.Builder> formatter)

Creates a new Gemini chat model instance.

参数:

apiKey - the API key for authentication (for Gemini API)

baseUrl - the custom base URL for Gemini API (null for default)

modelName - the model name to use (e.g., "gemini-2.0-flash", "gemini-1.5-pro")

streamEnabled - whether streaming should be enabled

project - the Google Cloud project ID (for Vertex AI)

location - the Google Cloud location (for Vertex AI, e.g., "us-central1")

vertexAI - whether to use Vertex AI APIs (null for auto-detection)

httpOptions - HTTP options for the client

credentials - Google credentials (for Vertex AI)

clientOptions - client options for the API client

defaultOptions - default generation options

formatter - the message formatter to use (null for default Gemini formatter)

GeminiChatModel

public GeminiChatModel(String apiKey,
 String modelName,
 boolean streamEnabled,
 String project,
 String location,
 Boolean vertexAI,
 com.google.genai.types.HttpOptions httpOptions,
 com.google.auth.oauth2.GoogleCredentials credentials,
 com.google.genai.types.ClientOptions clientOptions,
 GenerateOptions defaultOptions,
 Formatter<com.google.genai.types.Content,com.google.genai.types.GenerateContentResponse,com.google.genai.types.GenerateContentConfig.Builder> formatter)

Creates a new Gemini chat model instance using the default endpoint configuration.

This overload passes null for baseUrl, allowing the SDK to use its default endpoint behavior for either Gemini API or Vertex AI, depending on the provided configuration.

参数:

apiKey - the API key for authentication (for Gemini API)

modelName - the model name to use (e.g., "gemini-2.0-flash", "gemini-1.5-pro")

streamEnabled - whether streaming should be enabled

project - the Google Cloud project ID (for Vertex AI)

location - the Google Cloud location (for Vertex AI, e.g., "us-central1")

vertexAI - whether to use Vertex AI APIs (null for auto-detection)

httpOptions - HTTP options for the client

credentials - Google credentials (for Vertex AI)

clientOptions - client options for the API client

defaultOptions - default generation options

formatter - the message formatter to use (null for default Gemini formatter)

方法详细资料

doStream

protected reactor.core.publisher.Flux<ChatResponse> doStream(List<Msg> messages,
 List<ToolSchema> tools,
 GenerateOptions options)

Stream chat completion responses from Gemini's API.

This method internally handles message formatting using the configured formatter. When streaming is enabled, it returns incremental responses as they arrive. When streaming is disabled, it returns a single complete response.

指定者:

doStream 在类中 ChatModelBase

参数:

messages - AgentScope messages to send to the model

tools - Optional list of tool schemas (null or empty if no tools)

options - Optional generation options (null to use defaults)

返回:

Flux stream of chat responses

getModelName

public String getModelName()

从接口复制的说明: Model

Get model name for logging and identification.

返回:

model name

close

public void close()

Close the Gemini client.

builder

public static GeminiChatModel.Builder builder()

Creates a new builder for GeminiChatModel.

返回:

a new Builder instance