tessl/maven-dev-langchain4j--langchain4j-google-ai-gemini

LangChain4j integration for Google AI Gemini models providing chat, streaming, embeddings, image generation, and batch processing capabilities

Overview

Eval results

Files

LangChain4j Google AI Gemini

Name: tessl/maven-dev-langchain4j--langchain4j-google-ai-gemini
Author: tessl

A comprehensive Java integration library for Google AI Gemini models within the LangChain4j framework, enabling developers to build advanced AI-powered applications with enterprise Java frameworks. This package provides unified APIs for chat interactions (synchronous and streaming), embeddings, image generation, token estimation, file management, and cost-efficient batch processing.

Package Information

Package Name: langchain4j-google-ai-gemini
Group ID: dev.langchain4j
Artifact ID: langchain4j-google-ai-gemini
Version: 1.11.0
Language: Java
Package Manager: Maven
Installation:
```
<dependency>
    <groupId>dev.langchain4j</groupId>
    <artifactId>langchain4j-google-ai-gemini</artifactId>
    <version>1.11.0</version>
</dependency>
```

Core Imports

import dev.langchain4j.model.googleai.GoogleAiGeminiChatModel;
import dev.langchain4j.model.googleai.GoogleAiGeminiStreamingChatModel;
import dev.langchain4j.model.googleai.GoogleAiEmbeddingModel;
import dev.langchain4j.model.googleai.GoogleAiGeminiImageModel;
import dev.langchain4j.model.googleai.GeminiFiles;
import dev.langchain4j.model.googleai.GoogleAiGeminiTokenCountEstimator;

Basic Usage

import dev.langchain4j.model.googleai.GoogleAiGeminiChatModel;
import dev.langchain4j.data.message.ChatMessage;
import dev.langchain4j.model.output.Response;

// Create a chat model
GoogleAiGeminiChatModel model = GoogleAiGeminiChatModel.builder()
    .apiKey(System.getenv("GOOGLE_AI_API_KEY"))
    .modelName("gemini-2.5-pro")
    .temperature(0.7)
    .build();

// Send a chat message
Response<AiMessage> response = model.generate("What is the capital of France?");
System.out.println(response.content().text());

// Use streaming for real-time responses
GoogleAiGeminiStreamingChatModel streamingModel = GoogleAiGeminiStreamingChatModel.builder()
    .apiKey(System.getenv("GOOGLE_AI_API_KEY"))
    .modelName("gemini-2.5-flash")
    .build();

streamingModel.generate("Tell me a story", new StreamingResponseHandler<AiMessage>() {
    @Override
    public void onNext(String token) {
        System.out.print(token);
    }

    @Override
    public void onComplete(Response<AiMessage> response) {
        System.out.println("\nDone!");
    }

    @Override
    public void onError(Throwable error) {
        error.printStackTrace();
    }
});

Architecture

The langchain4j-google-ai-gemini package is organized around several key components:

Chat Models: Synchronous (GoogleAiGeminiChatModel) and streaming (GoogleAiGeminiStreamingChatModel) interfaces for conversational AI, with support for multimodal inputs (text, images, video, audio, PDFs), function calling, structured outputs, and thinking mode
Batch Processing: Cost-efficient batch operations (GoogleAiGeminiBatchChatModel, GoogleAiGeminiBatchEmbeddingModel, GoogleAiGeminiBatchImageModel) offering 50% cost reduction with 24-hour processing SLO
Embedding Models: Text embedding generation (GoogleAiEmbeddingModel) with configurable task types and output dimensionality
Image Models: Image generation and editing capabilities (GoogleAiGeminiImageModel) with safety settings and Google Search grounding
File Management: Gemini Files API integration (GeminiFiles) supporting up to 20GB project capacity and 2GB per-file uploads with 48-hour persistence
Utilities: Token counting (GoogleAiGeminiTokenCountEstimator) and model catalog (GoogleAiGeminiModelCatalog) for resource planning and model discovery
Configuration: Extensive configuration options through builder patterns including safety settings, function calling, response formats, and media resolution

All models follow the Builder Pattern for construction and integrate seamlessly with LangChain4j core interfaces, enabling use with Spring Boot, Quarkus, Helidon, and Micronaut frameworks.

Capabilities

Chat Models - Synchronous

Synchronous chat model for Gemini API supporting multimodal inputs, function calling, structured outputs, code execution, and thinking mode.

public class GoogleAiGeminiChatModel {
    public static GoogleAiGeminiChatModelBuilder builder();
    public ChatRequestParameters defaultRequestParameters();
    public ChatResponse doChat(ChatRequest chatRequest);
    public Set<Capability> supportedCapabilities();
    public List<ChatModelListener> listeners();
    public ModelProvider provider();
}

Chat Models - Synchronous

Chat Models - Streaming

Streaming chat model for Gemini API enabling real-time token-by-token responses with multimodal support and function calling.

public class GoogleAiGeminiStreamingChatModel {
    public static GoogleAiGeminiStreamingChatModelBuilder builder();
    public ChatRequestParameters defaultRequestParameters();
    public void doChat(ChatRequest request, StreamingChatResponseHandler handler);
    public List<ChatModelListener> listeners();
    public ModelProvider provider();
}

Chat Models - Streaming

Chat Models - Batch Processing

Batch processing for chat requests offering 50% cost reduction with 24-hour processing SLO (experimental).

public class GoogleAiGeminiBatchChatModel {
    public static Builder builder();
    public BatchResponse<ChatResponse> createBatchInline(String displayName, Long priority, List<ChatRequest> requests);
    public BatchResponse<ChatResponse> createBatchFromFile(String displayName, GeminiFile file);
    public void writeBatchToFile(JsonLinesWriter writer, Iterable<BatchFileRequest<ChatRequest>> requests);
    public BatchResponse<ChatResponse> retrieveBatchResults(BatchName name);
    public void cancelBatchJob(BatchName name);
    public void deleteBatchJob(BatchName name);
    public BatchList<ChatResponse> listBatchJobs(Integer pageSize, String pageToken);
}

Batch Processing

Embedding Models

Text embedding generation with configurable task types and output dimensionality.

public class GoogleAiEmbeddingModel {
    public static GoogleAiEmbeddingModelBuilder builder();
    public Response<Embedding> embed(TextSegment textSegment);
    public Response<Embedding> embed(String text);
    public Response<List<Embedding>> embedAll(List<TextSegment> textSegments);
    public String modelName();
    public Integer knownDimension();

    public enum TaskType {
        RETRIEVAL_QUERY,
        RETRIEVAL_DOCUMENT,
        SEMANTIC_SIMILARITY,
        CLASSIFICATION,
        CLUSTERING,
        QUESTION_ANSWERING,
        FACT_VERIFICATION
    }
}

Embedding Models

Image Models

Image generation and editing with safety settings and Google Search grounding (experimental). Includes utility class for extracting generated images from responses.

public class GoogleAiGeminiImageModel {
    public static GoogleAiGeminiImageModelBuilder builder();
    public String modelName();
    public Response<Image> generate(String prompt);
    public Response<Image> edit(Image image, String prompt);
    public Response<Image> edit(Image image, Image mask, String prompt);
}

public class GeneratedImageHelper {
    public static List<Image> getGeneratedImages(AiMessage aiMessage);
    public static boolean hasGeneratedImages(AiMessage aiMessage);
}

Image Models

File Management

Gemini Files API for uploading and managing files (images, videos, audio, PDFs) with up to 20GB project capacity and 2GB per-file limit.

public class GeminiFiles {
    public static Builder builder();
    public GeminiFile uploadFile(Path filePath, String displayName);
    public GeminiFile uploadFile(byte[] fileBytes, String mimeType, String name);
    public GeminiFile getMetadata(String name);
    public List<GeminiFile> listFiles();
    public void deleteFile(String name);
}

public record GeminiFile(
    String name,
    String displayName,
    String mimeType,
    Long sizeBytes,
    String createTime,
    String updateTime,
    String expirationTime,
    String sha256Hash,
    String uri,
    String state
) {
    public boolean isActive();
    public boolean isProcessing();
    public boolean isFailed();
}

File Management

Token Counting

Token count estimation for text, messages, and tool specifications to manage API costs and context limits.

public class GoogleAiGeminiTokenCountEstimator {
    public static Builder builder();
    public int estimateTokenCountInText(String text);
    public int estimateTokenCountInMessage(ChatMessage message);
    public int estimateTokenCountInMessages(Iterable<ChatMessage> messages);
    public int estimateTokenCountInToolExecutionRequests(Iterable<ToolExecutionRequest> requests);
    public int estimateTokenCountInToolSpecifications(Iterable<ToolSpecification> specifications);
}

Token Counting

Model Catalog

List available Gemini models and their capabilities.

public class GoogleAiGeminiModelCatalog {
    public static Builder builder();
    public List<ModelDescription> listModels();
    public ModelProvider provider();
}

Model Catalog

Configuration - Safety Settings

Content safety filtering configuration to control harmful content blocking.

public class GeminiSafetySetting {
    public GeminiSafetySetting(GeminiHarmCategory category, GeminiHarmBlockThreshold threshold);
    public GeminiHarmCategory getCategory();
    public GeminiHarmBlockThreshold getThreshold();
    public void setCategory(GeminiHarmCategory category);
    public void setThreshold(GeminiHarmBlockThreshold threshold);
}

public enum GeminiHarmCategory {
    HARM_CATEGORY_HATE_SPEECH,
    HARM_CATEGORY_SEXUALLY_EXPLICIT,
    HARM_CATEGORY_DANGEROUS_CONTENT,
    HARM_CATEGORY_HARASSMENT,
    HARM_CATEGORY_CIVIC_INTEGRITY
}

public enum GeminiHarmBlockThreshold {
    HARM_BLOCK_THRESHOLD_UNSPECIFIED,
    BLOCK_LOW_AND_ABOVE,
    BLOCK_MEDIUM_AND_ABOVE,
    BLOCK_ONLY_HIGH,
    BLOCK_NONE
}

Configuration

Advanced Configuration - Generation Control

Fine-grained control over generation behavior including structured output schemas, stop sequences, penalties, and response modalities.

public record GeminiGenerationConfig(
    List<String> stopSequences,
    String responseMimeType,
    Object responseSchema,
    Map<String, Object> responseJsonSchema,
    Integer candidateCount,
    Integer maxOutputTokens,
    Double temperature,
    Integer topK,
    Integer seed,
    Double topP,
    Double presencePenalty,
    Double frequencyPenalty,
    GeminiThinkingConfig thinkingConfig,
    Boolean responseLogprobs,
    Boolean enableEnhancedCivicAnswers,
    List<GeminiResponseModality> responseModalities,
    GeminiImageConfig imageConfig,
    Integer logprobs,
    GeminiMediaResolutionLevel mediaResolution
) {
    public static Builder builder();
}

Advanced Configuration

Response Metadata

Chat response metadata including grounding information from web search, Google Maps, and retrieval operations.

public class GoogleAiGeminiChatResponseMetadata {
    public static Builder builder();
    public GroundingMetadata groundingMetadata();
    public Builder toBuilder();
}

public record GroundingMetadata(
    List<GroundingChunk> groundingChunks,
    List<GroundingSupport> groundingSupports,
    List<String> webSearchQueries,
    SearchEntryPoint searchEntryPoint,
    RetrievalMetadata retrievalMetadata,
    String googleMapsWidgetContextToken
) {
    public static Builder builder();
}

Response Metadata

Batch Processing Types

Sealed interfaces and records for batch job management and result retrieval.

public sealed interface BatchResponse<T> permits BatchIncomplete, BatchSuccess, BatchError {
    // Permitted implementations in sub-doc
}

public record BatchList<T>(String pageToken, List<BatchResponse<T>> responses) {}

public record BatchName(String value) {}

public record BatchFileRequest<REQ>(String key, REQ request) {}

Batch Processing

JSON Lines Writing

Interface for writing batch request files in JSON Lines format.

public interface JsonLinesWriter {
    void write(Object object);
    void write(Iterable<?> objects);
    void flush();
    void close();
}

public class JsonLinesWriters {
    public static JsonLinesWriter streaming(Path path);
    public static JsonLinesWriter streaming(OutputStream outputStream);
}

Batch Processing

Install with Tessl CLI

npx tessl i tessl/maven-dev-langchain4j--langchain4j-google-ai-gemini

docs

advanced-configuration.md

tessl/maven-dev-langchain4j--langchain4j-google-ai-gemini

index.md.css-3qkkll{font-size:var(--chakra-font-sizes-sm);font-weight:var(--chakra-font-weights-normal);color:var(--chakra-colors-gray-300);}docs/

LangChain4j Google AI Gemini

Package Information

Core Imports

Basic Usage

Architecture

Capabilities

Chat Models - Synchronous

Chat Models - Streaming

Chat Models - Batch Processing

Embedding Models

Image Models

File Management

Token Counting

Model Catalog

Configuration - Safety Settings

Advanced Configuration - Generation Control

Response Metadata

Batch Processing Types

JSON Lines Writing

index.mddocs/