Aws Sdk Java V2 Bedrock

AWS SDK for Java 2.x - Amazon Bedrock

Overview

Invokes foundation models through AWS SDK for Java 2.x. Configures clients, builds model-specific JSON payloads, handles streaming responses with error recovery, creates embeddings for RAG, integrates generative AI into Spring Boot applications, and implements exponential backoff for resilience.

When to Use

• Invoke Claude, Llama, Titan, or Stable Diffusion for text/image generation
• Configure BedrockClient and BedrockRuntimeClient instances
• Build and parse model-specific payloads (Claude, Titan, Llama formats)
• Stream real-time AI responses with async handlers and error recovery
• Create embeddings for retrieval-augmented generation
• Integrate generative AI into Spring Boot microservices
• Handle throttling with exponential backoff retry logic

Quick Start

Dependencies

<!-- Bedrock (model management) -->
<dependency>
    <groupId>software.amazon.awssdk</groupId>
    <artifactId>bedrock</artifactId>
</dependency>

<!-- Bedrock Runtime (model invocation) -->
<dependency>
    <groupId>software.amazon.awssdk</groupId>
    <artifactId>bedrockruntime</artifactId>
</dependency>

<!-- For JSON processing -->
<dependency>
    <groupId>org.json</groupId>
    <artifactId>json</artifactId>
    <version>20231013</version>
</dependency>

Client Setup

import software.amazon.awssdk.regions.Region;
import software.amazon.awssdk.services.bedrock.BedrockClient;
import software.amazon.awssdk.services.bedrockruntime.BedrockRuntimeClient;

// Model management client
BedrockClient bedrockClient = BedrockClient.builder()
    .region(Region.US_EAST_1)
    .build();

// Model invocation client
BedrockRuntimeClient bedrockRuntimeClient = BedrockRuntimeClient.builder()
    .region(Region.US_EAST_1)
    .build();

Instructions

Follow these steps for production-ready Bedrock integration:

1. Configure AWS Credentials - Set up IAM roles with Bedrock permissions (avoid access keys) 2. Enable Model Access - Request access to specific foundation models in AWS Console 3. Initialize Clients - Create reusable BedrockClient and BedrockRuntimeClient instances 4. Validate Model Availability - Test with a simple invocation before production use 5. Build Payloads - Create model-specific JSON payloads with proper format 6. Handle Responses - Parse response structure and extract content 7. Implement Streaming - Use response stream handlers for real-time generation 8. Add Error Handling - Implement retry logic with exponential backoff

Validation Checkpoint: Always test with a simple prompt (e.g., "Hello") before production use to verify model access and response parsing.

Examples

Text Generation with Claude

public String generateWithClaude(BedrockRuntimeClient client, String prompt) {
    JSONObject payload = new JSONObject()
        .put("anthropic_version", "bedrock-2023-05-31")
        .put("max_tokens", 1000)
        .put("messages", new JSONObject[]{
            new JSONObject().put("role", "user").put("content", prompt)
        });

    InvokeModelResponse response = client.invokeModel(InvokeModelRequest.builder()
        .modelId("anthropic.claude-sonnet-4-5-20250929-v1:0")
        .body(SdkBytes.fromUtf8String(payload.toString()))
        .build());

    JSONObject responseBody = new JSONObject(response.body().asUtf8String());
    return responseBody.getJSONArray("content")
        .getJSONObject(0)
        .getString("text");
}

Model Discovery

import software.amazon.awssdk.services.bedrock.model.*;

public List<FoundationModelSummary> listFoundationModels(BedrockClient bedrockClient) {
    return bedrockClient.listFoundationModels().modelSummaries();
}

Multi-Model Invocation

public String invokeModel(BedrockRuntimeClient client, String modelId, String prompt) {
    JSONObject payload = createPayload(modelId, prompt);

    InvokeModelResponse response = client.invokeModel(request -> request
        .modelId(modelId)
        .body(SdkBytes.fromUtf8String(payload.toString())));

    return extractTextFromResponse(modelId, response.body().asUtf8String());
}

private JSONObject createPayload(String modelId, String prompt) {
    if (modelId.startsWith("anthropic.claude")) {
        return new JSONObject()
            .put("anthropic_version", "bedrock-2023-05-31")
            .put("max_tokens", 1000)
            .put("messages", new JSONObject[]{
                new JSONObject().put("role", "user").put("content", prompt)
            });
    } else if (modelId.startsWith("amazon.titan")) {
        return new JSONObject()
            .put("inputText", prompt)
            .put("textGenerationConfig", new JSONObject()
                .put("maxTokenCount", 512)
                .put("temperature", 0.7));
    } else if (modelId.startsWith("meta.llama")) {
        return new JSONObject()
            .put("prompt", "[INST] " + prompt + " [/INST]")
            .put("max_gen_len", 512)
            .put("temperature", 0.7);
    }
    throw new IllegalArgumentException("Unsupported model: " + modelId);
}

Streaming Response with Error Handling

public String streamResponseWithRetry(BedrockRuntimeClient client, String modelId, String prompt, int maxRetries) {
    int attempt = 0;
    while (attempt < maxRetries) {
        try {
            JSONObject payload = createPayload(modelId, prompt);
            StringBuilder fullResponse = new StringBuilder();

            InvokeModelWithResponseStreamRequest request = InvokeModelWithResponseStreamRequest.builder()
                .modelId(modelId)
                .body(SdkBytes.fromUtf8String(payload.toString()))
                .build();

            client.invokeModelWithResponseStream(request,
                InvokeModelWithResponseStreamResponseHandler.builder()
                    .onEventStream(stream -> stream.forEach(event -> {
                        if (event instanceof PayloadPart) {
                            String chunk = ((PayloadPart) event).bytes().asUtf8String();
                            fullResponse.append(chunk);
                        }
                    }))
                    .onError(e -> System.err.println("Stream error: " + e.getMessage()))
                    .build());

            return fullResponse.toString();
        } catch (Exception e) {
            attempt++;
            if (attempt >= maxRetries) {
                throw new RuntimeException("Stream failed after " + maxRetries + " attempts", e);
            }
            try {
                Thread.sleep((long) Math.pow(2, attempt) * 1000); // Exponential backoff
            } catch (InterruptedException ie) {
                Thread.currentThread().interrupt();
                throw new RuntimeException("Interrupted during retry", ie);
            }
        }
    }
    throw new RuntimeException("Unexpected error in streaming");
}

Exponential Backoff for Throttling

import software.amazon.awssdk.awscore.exception.AwsServiceException;

public <T> T invokeWithRetry(Supplier<T> invocation, int maxRetries) {
    int attempt = 0;
    while (attempt < maxRetries) {
        try {
            return invocation.get();
        } catch (AwsServiceException e) {
            if (e.statusCode() == 429 || e.statusCode() >= 500) {
                attempt++;
                if (attempt >= maxRetries) throw e;
                long delayMs = Math.min(1000 * (1L << attempt) + (long) (Math.random() * 1000), 30000);
                Thread.sleep(delayMs);
            } else {
                throw e;
            }
        }
    }
    throw new IllegalStateException("Should not reach here");
}

Text Embeddings

public double[] createEmbeddings(BedrockRuntimeClient client, String text) {
    String modelId = "amazon.titan-embed-text-v1";

    JSONObject payload = new JSONObject().put("inputText", text);

    InvokeModelResponse response = client.invokeModel(request -> request
        .modelId(modelId)
        .body(SdkBytes.fromUtf8String(payload.toString())));

    JSONObject responseBody = new JSONObject(response.body().asUtf8String());
    JSONArray embeddingArray = responseBody.getJSONArray("embedding");

    double[] embeddings = new double[embeddingArray.length()];
    for (int i = 0; i < embeddingArray.length(); i++) {
        embeddings[i] = embeddingArray.getDouble(i);
    }
    return embeddings;
}

Spring Boot Integration

@Configuration
public class BedrockConfiguration {

    @Bean
    public BedrockClient bedrockClient() {
        return BedrockClient.builder()
            .region(Region.US_EAST_1)
            .build();
    }

    @Bean
    public BedrockRuntimeClient bedrockRuntimeClient() {
        return BedrockRuntimeClient.builder()
            .region(Region.US_EAST_1)
            .build();
    }
}

@Service
public class BedrockAIService {

    private final BedrockRuntimeClient bedrockRuntimeClient;
    private final ObjectMapper mapper;

    @Value("${bedrock.default-model-id:anthropic.claude-sonnet-4-5-20250929-v1:0}")
    private String defaultModelId;

    public BedrockAIService(BedrockRuntimeClient bedrockRuntimeClient, ObjectMapper mapper) {
        this.bedrockRuntimeClient = bedrockRuntimeClient;
        this.mapper = mapper;
    }

    public String generateText(String prompt) {
        Map<String, Object> payload = Map.of(
            "anthropic_version", "bedrock-2023-05-31",
            "max_tokens", 1000,
            "messages", List.of(Map.of("role", "user", "content", prompt))
        );

        InvokeModelResponse response = bedrockRuntimeClient.invokeModel(
            InvokeModelRequest.builder()
                .modelId(defaultModelId)
                .body(SdkBytes.fromUtf8String(mapper.writeValueAsString(payload)))
                .build());

        return extractText(response.body().asUtf8String());
    }
}

See examples directory for comprehensive usage patterns.

Best Practices

Model Selection

• Claude 4.5 Sonnet: Complex reasoning, analysis, and creative tasks
• Claude 4.5 Haiku: Fast and affordable for real-time applications
• Llama 3.1: Open-source alternative for general tasks
• Titan: AWS native, cost-effective for simple text generation

Performance

• Reuse client instances (avoid creating new clients per request)
• Use async clients for I/O operations
• Implement streaming for long responses
• Cache foundation model lists

Security

• Never log sensitive prompt data
• Use IAM roles for authentication
• Sanitize user inputs to prevent prompt injection
• Implement rate limiting for public applications

Constraints and Warnings

• Cost Management: Bedrock API calls incur charges per token; implement usage monitoring and budget alerts.
• Model Access: Foundation models must be enabled in AWS Console; verify region availability.
• Rate Limits: Implement exponential backoff for throttling; check per-model limits.
• Payload Size: Maximum payload size varies by model; use chunking for large documents.
• Streaming Complexity: Handle partial content and error recovery carefully.
• Data Privacy: Prompts and responses may be logged by AWS; review data policies.
• Credentials: Never embed credentials in code; use IAM roles for EC2/Lambda.

Common Model IDs

• Claude Sonnet 4.5: anthropic.claude-sonnet-4-5-20250929-v1:0
• Claude Haiku 4.5: anthropic.claude-haiku-4-5-20251001-v1:0
• Llama 3.1 70B: meta.llama3-1-70b-instruct-v1:0
• Titan Embeddings: amazon.titan-embed-text-v1

See Model Reference for complete list.

References

• Advanced Topics - Multi-model patterns, advanced error handling
• Model Reference - Detailed specifications, payload formats
• Testing Strategies - Unit testing, LocalStack integration
• AWS Bedrock User Guide
• AWS SDK Examples
• Supported Models

Related Skills

• aws-sdk-java-v2-core - Core AWS SDK patterns
• langchain4j-ai-services-patterns - LangChain4j integration
• spring-boot-dependency-injection - Spring DI patterns

Advanced Model Patterns

Model-Specific Configuration

Claude Models Configuration

// Claude 3 Sonnet
public String invokeClaude3Sonnet(BedrockRuntimeClient client, String prompt) {
    String modelId = "anthropic.claude-3-sonnet-20240229-v1:0";

    JSONObject payload = new JSONObject()
        .put("anthropic_version", "bedrock-2023-05-31")
        .put("max_tokens", 1000)
        .put("temperature", 0.7)
        .put("top_p", 1.0)
        .put("messages", new JSONObject[]{
            new JSONObject()
                .put("role", "user")
                .put("content", prompt)
        });

    InvokeModelResponse response = client.invokeModel(request -> request
        .modelId(modelId)
        .body(SdkBytes.fromUtf8String(payload.toString())));

    JSONObject responseBody = new JSONObject(response.body().asUtf8String());
    return responseBody.getJSONArray("content")
        .getJSONObject(0)
        .getString("text");
}

// Claude 3 Haiku (faster, cheaper)
public String invokeClaude3Haiku(BedrockRuntimeClient client, String prompt) {
    String modelId = "anthropic.claude-3-haiku-20240307-v1:0";

    JSONObject payload = new JSONObject()
        .put("anthropic_version", "bedrock-2023-05-31")
        .put("max_tokens", 400)
        .put("messages", new JSONObject[]{
            new JSONObject()
                .put("role", "user")
                .put("content", prompt)
        });

    // Similar invocation pattern as above
}

Llama Models Configuration

// Llama 3 70B
public String invokeLlama3_70B(BedrockRuntimeClient client, String prompt) {
    String modelId = "meta.llama3-70b-instruct-v1:0";

    JSONObject payload = new JSONObject()
        .put("prompt", prompt)
        .put("max_gen_len", 512)
        .put("temperature", 0.7)
        .put("top_p", 0.9)
        .put("stop", new String[]{"[INST]", "[/INST]"}); // Custom stop tokens

    InvokeModelResponse response = client.invokeModel(request -> request
        .modelId(modelId)
        .body(SdkBytes.fromUtf8String(payload.toString())));

    JSONObject responseBody = new JSONObject(response.body().asUtf8String());
    return responseBody.getString("generation");
}

Multi-Model Service Layer

@Service
public class MultiModelService {

    private final BedrockRuntimeClient bedrockRuntimeClient;
    private final ObjectMapper objectMapper;

    public MultiModelService(BedrockRuntimeClient bedrockRuntimeClient,
                           ObjectMapper objectMapper) {
        this.bedrockRuntimeClient = bedrockRuntimeClient;
        this.objectMapper = objectMapper;
    }

    public String invokeModel(String modelId, String prompt, Map<String, Object> additionalParams) {
        Map<String, Object> payload = createModelPayload(modelId, prompt, additionalParams);

        try {
            InvokeModelResponse response = bedrockRuntimeClient.invokeModel(
                request -> request
                    .modelId(modelId)
                    .body(SdkBytes.fromUtf8String(objectMapper.writeValueAsString(payload))));

            return extractResponseContent(modelId, response.body().asUtf8String());

        } catch (Exception e) {
            throw new RuntimeException("Model invocation failed: " + e.getMessage(), e);
        }
    }

    private Map<String, Object> createModelPayload(String modelId, String prompt,
                                                   Map<String, Object> additionalParams) {
        Map<String, Object> payload = new HashMap<>();

        if (modelId.startsWith("anthropic.claude")) {
            payload.put("anthropic_version", "bedrock-2023-05-31");
            payload.put("messages", List.of(Map.of("role", "user", "content", prompt)));

            // Add common parameters with defaults
            payload.putIfAbsent("max_tokens", 1000);
            payload.putIfAbsent("temperature", 0.7);

        } else if (modelId.startsWith("meta.llama")) {
            payload.put("prompt", prompt);
            payload.putIfAbsent("max_gen_len", 512);
            payload.putIfAbsent("temperature", 0.7);

        } else if (modelId.startsWith("amazon.titan")) {
            payload.put("inputText", prompt);
            payload.putIfAbsent("textGenerationConfig",
                Map.of("maxTokenCount", 512, "temperature", 0.7));
        }

        // Add additional parameters
        if (additionalParams != null) {
            payload.putAll(additionalParams);
        }

        return payload;
    }
}

Advanced Error Handling

@Component
public class BedrockErrorHandler {

    @Retryable(value = {SdkClientException.class}, maxAttempts = 3, backoff = @Backoff(delay = 1000))
    public String invokeWithRetry(BedrockRuntimeClient client, String modelId,
                                String payloadJson) {
        try {
            InvokeModelResponse response = client.invokeModel(request -> request
                .modelId(modelId)
                .body(SdkBytes.fromUtf8String(payloadJson)));
            return response.body().asUtf8String();

        } catch (ThrottlingException e) {
            // Exponential backoff for throttling
            throw new RuntimeException("Rate limit exceeded, please try again later", e);
        } catch (ValidationException e) {
            throw new IllegalArgumentException("Invalid request: " + e.getMessage(), e);
        } catch (SdkException e) {
            throw new RuntimeException("AWS SDK error: " + e.getMessage(), e);
        }
    }
}

Batch Processing

@Service
public class BedrockBatchService {

    public List<String> processBatch(BedrockRuntimeClient client, String modelId,
                                    List<String> prompts) {
        return prompts.parallelStream()
            .map(prompt -> invokeModelWithTimeout(client, modelId, prompt, 30))
            .collect(Collectors.toList());
    }

    private String invokeModelWithTimeout(BedrockRuntimeClient client, String modelId,
                                         String prompt, int timeoutSeconds) {
        ExecutorService executor = Executors.newSingleThreadExecutor();
        Future<String> future = executor.submit(() -> {
            JSONObject payload = new JSONObject()
                .put("prompt", prompt)
                .put("max_tokens", 500);

            InvokeModelResponse response = client.invokeModel(request -> request
                .modelId(modelId)
                .body(SdkBytes.fromUtf8String(payload.toString())));

            return response.body().asUtf8String();
        });

        try {
            return future.get(timeoutSeconds, TimeUnit.SECONDS);
        } catch (TimeoutException e) {
            future.cancel(true);
            throw new RuntimeException("Model invocation timed out");
        } catch (Exception e) {
            throw new RuntimeException("Batch processing error", e);
        } finally {
            executor.shutdown();
        }
    }
}

Model Performance Optimization

@Configuration
public class BedrockOptimizationConfig {

    @Bean
    public BedrockRuntimeClient optimizedBedrockRuntimeClient() {
        return BedrockRuntimeClient.builder()
            .region(Region.US_EAST_1)
            .overrideConfiguration(ClientOverrideConfiguration.builder()
                .apiCallTimeout(Duration.ofSeconds(30))
                .apiCallAttemptTimeout(Duration.ofSeconds(20))
                .build())
            .httpClient(ApacheHttpClient.builder()
                .connectionTimeout(Duration.ofSeconds(10))
                .socketTimeout(Duration.ofSeconds(30))
                .build())
            .build();
    }
}

Custom Response Parsing

public class BedrockResponseParser {

    public static TextResponse parseTextResponse(String modelId, String responseBody) {
        try {
            switch (getModelProvider(modelId)) {
                case ANTHROPIC:
                    return parseAnthropicResponse(responseBody);
                case META:
                    return parseMetaResponse(responseBody);
                case AMAZON:
                    return parseAmazonResponse(responseBody);
                default:
                    throw new IllegalArgumentException("Unsupported model: " + modelId);
            }
        } catch (Exception e) {
            throw new ResponseParsingException("Failed to parse response for model: " + modelId, e);
        }
    }

    private static TextResponse parseAnthropicResponse(String responseBody) throws JSONException {
        JSONObject json = new JSONObject(responseBody);
        JSONArray content = json.getJSONArray("content");
        String text = content.getJSONObject(0).getString("text");
        int usage = json.getJSONObject("usage").getInt("input_tokens");

        return new TextResponse(text, usage, "anthropic");
    }

    private static TextResponse parseMetaResponse(String responseBody) throws JSONException {
        JSONObject json = new JSONObject(responseBody);
        String text = json.getString("generation");
        // Note: Meta doesn't provide token usage in basic response

        return new TextResponse(text, 0, "meta");
    }

    private enum ModelProvider {
        ANTHROPIC, META, AMAZON
    }

    public record TextResponse(String content, int tokensUsed, String provider) {}
}

Advanced Amazon Bedrock Topics

This document covers advanced patterns and topics for working with Amazon Bedrock using AWS SDK for Java 2.x.

Multi-Model Service Pattern

Create a service that can handle multiple foundation models with unified interfaces.

@Service
public class MultiModelAIService {

    private final BedrockRuntimeClient bedrockRuntimeClient;

    public MultiModelAIService(BedrockRuntimeClient bedrockRuntimeClient) {
        this.bedrockRuntimeClient = bedrockRuntimeClient;
    }

    public GenerationResult generate(GenerationRequest request) {
        String modelId = request.getModelId();
        String prompt = request.getPrompt();

        switch (getModelProvider(modelId)) {
            case ANTHROPIC:
                return generateWithAnthropic(modelId, prompt, request.getConfig());
            case AMAZON:
                return generateWithAmazon(modelId, prompt, request.getConfig());
            case META:
                return generateWithMeta(modelId, prompt, request.getConfig());
            default:
                throw new IllegalArgumentException("Unsupported model provider: " + modelId);
        }
    }

    private GenerationProvider getModelProvider(String modelId) {
        if (modelId.startsWith("anthropic.")) return GenerationProvider.ANTHROPIC;
        if (modelId.startsWith("amazon.")) return GenerationProvider.AMazon;
        if (modelId.startsWith("meta.")) return GenerationProvider.META;
        throw new IllegalArgumentException("Unknown provider for model: " + modelId);
    }
}

Advanced Error Handling with Retries

Implement robust error handling with exponential backoff:

import software.amazon.awssdk.core.retry.RetryPolicy;
import software.amazon.awssdk.core.retry.backoff.BackoffStrategy;
import software.amazon.awssdk.core.retry.conditions.RetryCondition;
import software.amazon.awssdk.core.retry.predicates.RetryExceptionPredicates;

public class BedrockWithRetry {

    private final BedrockRuntimeClient client;
    private final RetryPolicy retryPolicy;

    public BedrockWithRetry(BedrockRuntimeClient client) {
        this.client = client;
        this.retryPolicy = RetryPolicy.builder()
            .numRetries(3)
            .retryCondition(RetryExceptionPredicates.equalTo(
                ThrottlingException.class))
            .backoffStrategy(BackoffStrategy.defaultStrategy())
            .build();
    }

    public String invokeModelWithRetry(String modelId, String payload) {
        try {
            InvokeModelRequest request = InvokeModelRequest.builder()
                .modelId(modelId)
                .body(SdkBytes.fromUtf8String(payload))
                .build();

            InvokeModelResponse response = client.invokeModel(request);
            return response.body().asUtf8String();

        } catch (ThrottlingException e) {
            throw new BedrockThrottledException("Rate limit exceeded for model: " + modelId, e);
        } catch (ValidationException e) {
            throw new BedrockValidationException("Invalid request for model: " + modelId, e);
        }
    }
}

Batch Processing Strategies

Process multiple requests efficiently:

@Service
public class BatchGenerationService {

    private final BedrockRuntimeClient bedrockRuntimeClient;

    public BatchGenerationService(BedrockRuntimeClient bedrockRuntimeClient) {
        this.bedrockRuntimeClient = bedrockRuntimeClient;
    }

    public List<BatchResult> processBatch(List<BatchRequest> requests) {
        // Process in parallel
        return requests.parallelStream()
            .map(this::processSingleRequest)
            .collect(Collectors.toList());
    }

    private BatchResult processSingleRequest(BatchRequest request) {
        try {
            InvokeModelRequest modelRequest = InvokeModelRequest.builder()
                .modelId(request.getModelId())
                .body(SdkBytes.fromUtf8String(request.getPayload()))
                .build();

            InvokeModelResponse response = bedrockRuntimeClient.invokeModel(modelRequest);

            return BatchResult.success(
                request.getRequestId(),
                response.body().asUtf8String()
            );

        } catch (Exception e) {
            return BatchResult.failure(request.getRequestId(), e.getMessage());
        }
    }
}

Performance Optimization Techniques

Connection Pooling

import software.amazon.awssdk.http.nio.apache.ApacheHttpClient;
import software.amazon.awssdk.http.apache.ProxyConfiguration;
import software.amazon.awssdk.regions.Region;

public class BedrockClientFactory {

    public static BedrockRuntimeClient createOptimizedClient() {
        ApacheHttpClient httpClient = ApacheHttpClient.builder()
            .connectionPoolMaxConnections(50)
            .socketTimeout(Duration.ofSeconds(30))
            .connectionTimeout(Duration.ofSeconds(30))
            .build();

        return BedrockRuntimeClient.builder()
            .region(Region.US_EAST_1)
            .httpClient(httpClient)
            .build();
    }
}

Response Caching

import com.github.benmanes.caffeine.cache.Cache;
import com.github.benmanes.caffeine.cache.Caffeine;

@Service
public class CachedAIService {

    private final BedrockRuntimeClient bedrockRuntimeClient;
    private final Cache<String, String> responseCache;

    public CachedAIService(BedrockRuntimeClient bedrockRuntimeClient) {
        this.bedrockRuntimeClient = bedrockRuntimeClient;
        this.responseCache = Caffeine.newBuilder()
            .maximumSize(1000)
            .expireAfterWrite(1, TimeUnit.HOURS)
            .build();
    }

    public String generateText(String prompt, String modelId) {
        String cacheKey = modelId + ":" + prompt.hashCode();

        return responseCache.get(cacheKey, key -> {
            String payload = createPayload(modelId, prompt);
            InvokeModelRequest request = InvokeModelRequest.builder()
                .modelId(modelId)
                .body(SdkBytes.fromUtf8String(payload))
                .build();

            InvokeModelResponse response = bedrockRuntimeClient.invokeModel(request);
            return response.body().asUtf8String();
        });
    }
}

Custom Response Parsing

Create specialized parsers for different model responses:

public interface ResponseParser {
    String parse(String responseJson);
}

public class AnthropicResponseParser implements ResponseParser {
    @Override
    public String parse(String responseJson) {
        try {
            JSONObject jsonResponse = new JSONObject(responseJson);
            return jsonResponse.getJSONArray("content")
                .getJSONObject(0)
                .getString("text");
        } catch (Exception e) {
            throw new ResponseParsingException("Failed to parse Anthropic response", e);
        }
    }
}

public class AmazonTitanResponseParser implements ResponseParser {
    @Override
    public String parse(String responseJson) {
        try {
            JSONObject jsonResponse = new JSONObject(responseJson);
            return jsonResponse.getJSONArray("results")
                .getJSONObject(0)
                .getString("outputText");
        } catch (Exception e) {
            throw new ResponseParsingException("Failed to parse Amazon Titan response", e);
        }
    }
}

public class LlamaResponseParser implements ResponseParser {
    @Override
    public String parse(String responseJson) {
        try {
            JSONObject jsonResponse = new JSONObject(responseJson);
            return jsonResponse.getString("generation");
        } catch (Exception e) {
            throw new ResponseParsingException("Failed to parse Llama response", e);
        }
    }
}

Metrics and Monitoring

Implement comprehensive monitoring:

import io.micrometer.core.instrument.MeterRegistry;
import io.micrometer.core.instrument.Timer;

@Service
public class MonitoredAIService {

    private final BedrockRuntimeClient bedrockRuntimeClient;
    private final Timer generationTimer;
    private final Counter errorCounter;

    public MonitoredAIService(BedrockRuntimeClient bedrockRuntimeClient,
                             MeterRegistry meterRegistry) {
        this.bedrockRuntimeClient = bedrockRuntimeClient;
        this.generationTimer = Timer.builder("bedrock.generation.time")
            .description("Time spent generating text with Bedrock")
            .register(meterRegistry);
        this.errorCounter = Counter.builder("bedrock.generation.errors")
            .description("Number of generation errors")
            .register(meterRegistry);
    }

    public String generateText(String prompt, String modelId) {
        return generationTimer.record(() -> {
            try {
                String payload = createPayload(modelId, prompt);
                InvokeModelRequest request = InvokeModelRequest.builder()
                    .modelId(modelId)
                    .body(SdkBytes.fromUtf8String(payload))
                    .build();

                InvokeModelResponse response = bedrockRuntimeClient.invokeModel(request);
                return response.body().asUtf8String();

            } catch (Exception e) {
                errorCounter.increment();
                throw new GenerationException("Failed to generate text", e);
            }
        });
    }
}

Advanced Configuration Management

@Configuration
@ConfigurationProperties(prefix = "bedrock")
public class AdvancedBedrockConfiguration {

    private String defaultRegion = "us-east-1";
    private int maxRetries = 3;
    private Duration timeout = Duration.ofSeconds(30);
    private boolean enableMetrics = true;
    private int maxCacheSize = 1000;
    private Duration cacheExpireAfter = Duration.ofHours(1);

    @Bean
    @Primary
    public BedrockRuntimeClient bedrockRuntimeClient() {
        BedrockRuntimeClient.Builder builder = BedrockRuntimeClient.builder()
            .region(Region.of(defaultRegion));

        if (enableMetrics) {
            builder.overrideConfiguration(c -> c.putAdvancedProperty(
                "metrics.enabled", "true"));
        }

        return builder.build();
    }

    // Getters and setters
}

Streaming Response Handling

Advanced streaming with proper backpressure handling:

@Service
public class StreamingAIService {

    private final BedrockRuntimeClient bedrockRuntimeClient;

    public StreamingAIService(BedrockRuntimeClient bedrockRuntimeClient) {
        this.bedrockRuntimeClient = bedrockRuntimeClient;
    }

    public Flux<String> streamResponse(String modelId, String prompt) {
        InvokeModelWithResponseStreamRequest request =
            InvokeModelWithResponseStreamRequest.builder()
                .modelId(modelId)
                .body(SdkBytes.fromUtf8String(createPayload(modelId, prompt)))
                .build();

        return Mono.fromCallable(() ->
            bedrockRuntimeClient.invokeModelWithResponseStream(request))
            .flatMapMany(responseStream -> Flux.defer(() ->
                Flux.create(sink -> {
                    responseStream.stream().forEach(event -> {
                        if (event instanceof PayloadPart) {
                            PayloadPart payloadPart = (PayloadPart) event;
                            String chunk = payloadPart.bytes().asUtf8String();
                            processChunk(chunk, sink);
                        }
                    });
                    sink.complete();
                }))
            )
            .onErrorResume(e -> Flux.error(new StreamingException("Stream failed", e)));
    }

    private void processChunk(String chunk, FluxSink<String> sink) {
        try {
            JSONObject chunkJson = new JSONObject(chunk);
            if (chunkJson.getString("type").equals("content_block_delta")) {
                String text = chunkJson.getJSONObject("delta").getString("text");
                sink.next(text);
            }
        } catch (Exception e) {
            sink.error(new ChunkProcessingException("Failed to process chunk", e));
        }
    }
}

<!DOCTYPE html> <!DOCTYPE HTML><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8"><title>Amazon Bedrock</title><meta xmlns="" name="subtitle" content="API Reference"><meta xmlns="" name="abstract" content="Details about operations and parameters in the Amazon Bedrock API Reference"><meta http-equiv="refresh" content="10;URL=welcome.html"><script type="text/javascript"><!-- var myDefaultPage = "welcome.html"; var myPage = document.location.search.substr(1); var myHash = document.location.hash;

if (myPage == null || myPage.length == 0) { myPage = myDefaultPage; } else { var docfile = myPage.match(/[^=\;\/?:\s]+\.html/); if (docfile == null) { myPage = myDefaultPage; } else { myPage = docfile + myHash; } } self.location.replace(myPage); --></script></head><body></body></html>

<!DOCTYPE html> <!DOCTYPE HTML><html xmlns="http://www.w3.org/1999/xhtml"><head><meta http-equiv="Content-Type" content="text/html; charset=UTF-8"><title>Amazon Bedrock</title><meta xmlns="" name="subtitle" content="User Guide"><meta xmlns="" name="abstract" content="User Guide for the Amazon Bedrock service."><meta http-equiv="refresh" content="10;URL=what-is-bedrock.html"><script type="text/javascript"><!-- var myDefaultPage = "what-is-bedrock.html"; var myPage = document.location.search.substr(1); var myHash = document.location.hash;

if (myPage == null || myPage.length == 0) { myPage = myDefaultPage; } else { var docfile = myPage.match(/[^=\;\/?:\s]+\.html/); if (docfile == null) { myPage = myDefaultPage; } else { myPage = docfile + myHash; } } self.location.replace(myPage); --></script></head><body></body></html>

<!DOCTYPE HTML> <html lang="en"> <head> <!-- Generated by javadoc (23) on Tue Oct 28 00:04:26 UTC 2025 --> <title>software.amazon.awssdk.services.bedrock (AWS SDK for Java - 2.36.3)</title> <meta name="viewport" content="width=device-width, initial-scale=1"> <meta http-equiv="Content-Type" content="text/html; charset=UTF-8"> <meta name="dc.created" content="2025-10-28"> <meta name="description" content="declaration: package: software.amazon.awssdk.services.bedrock"> <meta name="generator" content="javadoc/PackageWriter"> <link rel="stylesheet" type="text/css" href="../../../../../resource-files/jquery-ui.min.css" title="Style"> <link rel="stylesheet" type="text/css" href="../../../../../resource-files/stylesheet.css" title="Style"> <link rel="stylesheet" type="text/css" href="../../../../../resource-files/aws-sdk-java-v2-javadoc.css" title="Style"> <script type="text/javascript" src="../../../../../script-files/script.js"></script> <script type="text/javascript" src="../../../../../script-files/jquery-3.7.1.min.js"></script> <script type="text/javascript" src="../../../../../script-files/jquery-ui.min.js"></script> </head> <body class="package-declaration-page"> <script type="text/javascript">const pathtoroot = "../../../../../"; loadScripts(document, 'script');</script> <noscript> <div>JavaScript is disabled on your browser.</div> </noscript> <header role="banner"> <nav role="navigation"> <!-- ========= START OF TOP NAVBAR ======= --> <div class="top-nav" id="navbar-top"> <div class="nav-content"> <div class="nav-menu-button"><button id="navbar-toggle-button" aria-controls="navbar-top" aria-expanded="false" aria-label="Toggle navigation links"><span class="nav-bar-toggle-icon">&nbsp;</span><span class="nav-bar-toggle-icon">&nbsp;</span><span class="nav-bar-toggle-icon">&nbsp;</span></button></div> <div class="skip-nav"><a href="#skip-navbar-top" title="Skip navigation links">Skip navigation links</a></div> <ul id="navbar-top-firstrow" class="nav-list" title="Navigation"> <li><a href="../../../../../index.html">Overview</a></li> <li class="nav-bar-cell1-rev">Package</li> <li><a href="../../../../../index-all.html">Index</a></li> <li><a href="../../../../../search.html">Search</a></li> <li><a href="../../../../../help-doc.html#package">Help</a></li> </ul> <div class="about-language"><h2>AWS SDK for Java API Reference - 2.36.3</h2></div> </div> </div> <div class="sub-nav"> <div class="nav-content"> <ol class="sub-nav-list"> <li><a href="package-summary.html" class="current-selection">software.amazon.awssdk.services.bedrock</a></li> </ol> <div class="nav-list-search"> <input type="text" id="search-input" disabled placeholder="Search" aria-label="Search in documentation" autocomplete="off"> <input type="reset" id="reset-search" disabled value="Reset"> </div> </div> </div> <!-- ========= END OF TOP NAVBAR ========= --> <span class="skip-nav" id="skip-navbar-top"></span></nav> </header> <div class="main-grid"> <nav role="navigation" class="toc" aria-label="Table of contents"> <div class="toc-header">Contents</div> <button class="hide-sidebar"><span>Hide sidebar&nbsp;</span>&#10094;</button><button class="show-sidebar">&#10095;<span>&nbsp;Show sidebar</span></button> <ol class="toc-list"> <li><a href="#" tabindex="0">Description</a></li> <li><a href="#related-package-summary" tabindex="0">Related Packages</a></li> <li><a href="#class-summary" tabindex="0">Classes and Interfaces</a></li> </ol> </nav> <main role="main"> <div class="header"> <h1 title="Package software.amazon.awssdk.services.bedrock" class="title">Package software.amazon.awssdk.services.bedrock</h1> </div> <hr> <div class="horizontal-scroll"> <div class="package-signature">package <span class="element-name">software.amazon.awssdk.services.bedrock</span></div> <section class="package-description" id="package-description"> <div class="block"><p> Describes the API operations for creating, managing, fine-turning, and evaluating Amazon Bedrock models. </p></div> </section> </div> <section class="summary"> <ul class="summary-list"> <li> <div id="related-package-summary"> <div class="caption"><span>Related Packages</span></div> <div class="summary-table two-column-summary"> <div class="table-header col-first">Package</div> <div class="table-header col-last">Description</div> <div class="col-first even-row-color"><a href="endpoints/package-summary.html">software.amazon.awssdk.services.bedrock.endpoints</a></div> <div class="col-last even-row-color">&nbsp;</div> <div class="col-first odd-row-color"><a href="internal/package-summary.html">software.amazon.awssdk.services.bedrock.internal</a></div> <div class="col-last odd-row-color">&nbsp;</div> <div class="col-first even-row-color"><a href="model/package-summary.html">software.amazon.awssdk.services.bedrock.model</a></div> <div class="col-last even-row-color">&nbsp;</div> <div class="col-first odd-row-color"><a href="paginators/package-summary.html">software.amazon.awssdk.services.bedrock.paginators</a></div> <div class="col-last odd-row-color">&nbsp;</div> <div class="col-first even-row-color"><a href="transform/package-summary.html">software.amazon.awssdk.services.bedrock.transform</a></div> <div class="col-last even-row-color">&nbsp;</div> </div> </div> </li> <li> <div id="class-summary"> <div class="table-tabs" role="tablist" aria-orientation="horizontal"><button id="class-summary-tab0" role="tab" aria-selected="true" aria-controls="class-summary.tabpanel" tabindex="0" onkeydown="switchTab(event)" onclick="show('class-summary', 'class-summary', 2)" class="active-table-tab">All Classes and Interfaces</button><button id="class-summary-tab1" role="tab" aria-selected="false" aria-controls="class-summary.tabpanel" tabindex="-1" onkeydown="switchTab(event)" onclick="show('class-summary', 'class-summary-tab1', 2)" class="table-tab">Interfaces</button><button id="class-summary-tab2" role="tab" aria-selected="false" aria-controls="class-summary.tabpanel" tabindex="-1" onkeydown="switchTab(event)" onclick="show('class-summary', 'class-summary-tab2', 2)" class="table-tab">Classes</button></div> <div id="class-summary.tabpanel" role="tabpanel" aria-labelledby="class-summary-tab0"> <div class="summary-table two-column-summary"> <div class="table-header col-first">Class</div> <div class="table-header col-last">Description</div> <div class="col-first even-row-color class-summary class-summary-tab1"><a href="BedrockAsyncClient.html" title="interface in software.amazon.awssdk.services.bedrock">BedrockAsyncClient</a></div> <div class="col-last even-row-color class-summary class-summary-tab1"> <div class="block">Service client for accessing Amazon Bedrock asynchronously.</div> </div> <div class="col-first odd-row-color class-summary class-summary-tab1"><a href="BedrockAsyncClientBuilder.html" title="interface in software.amazon.awssdk.services.bedrock">BedrockAsyncClientBuilder</a></div> <div class="col-last odd-row-color class-summary class-summary-tab1"> <div class="block">A builder for creating an instance of <a href="BedrockAsyncClient.html" title="interface in software.amazon.awssdk.services.bedrock"><code>BedrockAsyncClient</code></a>.</div> </div> <div class="col-first even-row-color class-summary class-summary-tab1"><a href="BedrockBaseClientBuilder.html" title="interface in software.amazon.awssdk.services.bedrock">BedrockBaseClientBuilder</a>&lt;B extends <a href="BedrockBaseClientBuilder.html" title="interface in software.amazon.awssdk.services.bedrock">BedrockBaseClientBuilder</a>&lt;B,<wbr>C&gt;,<wbr>C&gt;</div> <div class="col-last even-row-color class-summary class-summary-tab1"> <div class="block">This includes configuration specific to Amazon Bedrock that is supported by both <a href="BedrockClientBuilder.html" title="interface in software.amazon.awssdk.services.bedrock"><code>BedrockClientBuilder</code></a> and <a href="BedrockAsyncClientBuilder.html" title="interface in software.amazon.awssdk.services.bedrock"><code>BedrockAsyncClientBuilder</code></a>.</div> </div> <div class="col-first odd-row-color class-summary class-summary-tab1"><a href="BedrockClient.html" title="interface in software.amazon.awssdk.services.bedrock">BedrockClient</a></div> <div class="col-last odd-row-color class-summary class-summary-tab1"> <div class="block">Service client for accessing Amazon Bedrock.</div> </div> <div class="col-first even-row-color class-summary class-summary-tab1"><a href="BedrockClientBuilder.html" title="interface in software.amazon.awssdk.services.bedrock">BedrockClientBuilder</a></div> <div class="col-last even-row-color class-summary class-summary-tab1"> <div class="block">A builder for creating an instance of <a href="BedrockClient.html" title="interface in software.amazon.awssdk.services.bedrock"><code>BedrockClient</code></a>.</div> </div> <div class="col-first odd-row-color class-summary class-summary-tab2"><a href="BedrockServiceClientConfiguration.html" title="class in software.amazon.awssdk.services.bedrock">BedrockServiceClientConfiguration</a></div> <div class="col-last odd-row-color class-summary class-summary-tab2"> <div class="block">Class to expose the service client settings to the user.</div> </div> <div class="col-first even-row-color class-summary class-summary-tab1"><a href="BedrockServiceClientConfiguration.Builder.html" title="interface in software.amazon.awssdk.services.bedrock">BedrockServiceClientConfiguration.Builder</a></div> <div class="col-last even-row-color class-summary class-summary-tab1"> <div class="block">A builder for creating a <a href="BedrockServiceClientConfiguration.html" title="class in software.amazon.awssdk.services.bedrock"><code>BedrockServiceClientConfiguration</code></a></div> </div> </div> </div> </div> </li> </ul> </section> <footer role="contentinfo"> <hr> <p class="legal-copy"><small><div style="margin:1.2em;"><h3><a id="fdbk" target="_blank">Provide feedback</a><h3></div> <span id="awsdocs-legal-zone-copyright"></span> <script type="text/javascript">document.addEventListener("DOMContentLoaded",()=>{ var a=document.createElement("meta"),b=document.createElement("meta"),c=document.createElement("script"), h=document.getElementsByTagName("head")[0],l=location.href,f=document.getElementById("fdbk"); a.name="guide-name",a.content="API Reference";b.name="service-name",b.content="AWS SDK for Java"; c.setAttribute("type","text/javascript"),c.setAttribute("src", "https://docs.aws.amazon.com/assets/js/awsdocs-boot.js");h.appendChild(a);h.appendChild(b); h.appendChild(c);f.setAttribute("href", "https://docs-feedback.aws.amazon.com/feedback.jsp?hidden_service_name="+ encodeURI("AWS SDK for Java")+"&topic_url="+encodeURI(l))}); </script></small></p> </footer> </main> </div> </body> </html>

Amazon Bedrock examples using SDK for Java 2.x - AWS SDK for Java 2.x

Amazon Bedrock examples using SDK for Java 2.x - AWS SDK for Java 2.x

[Open PDF](https://docs.aws.amazon.com/sdk-for-java/latest/developer-guide/java_bedrock_code_examples.html/pdfs/sdk-for-java/latest/developer-guide/aws-sdk-java-dg-v2.pdf#java_bedrock_code_examples "Open PDF")

Documentation AWS SDK for Java Developer Guide for version 2.x

Actions

Amazon Bedrock examples using SDK for Java 2.x

The following code examples show you how to perform actions and implement common scenarios by using the AWS SDK for Java 2.x with Amazon Bedrock.

_Actions_ are code excerpts from larger programs and must be run in context. While actions show you how to call individual service functions, you can see actions in context in their related scenarios.

Each example includes a link to the complete source code, where you can find instructions on how to set up and run the code in context.

Topics

• Actions

Actions

The following code example shows how to use GetFoundationModel.

SDK for Java 2.x

Note

There's more on GitHub. Find the complete example and learn how to set up and run in the AWS Code\ Examples Repository.

Get details about a foundation model using the synchronous Amazon Bedrock client.

    /**
     * Get details about an Amazon Bedrock foundation model.
     *
     * @param bedrockClient   The service client for accessing Amazon Bedrock.
     * @param modelIdentifier The model identifier.
     * @return An object containing the foundation model's details.
     */
    public static FoundationModelDetails getFoundationModel(BedrockClient bedrockClient, String modelIdentifier) {
        try {
            GetFoundationModelResponse response = bedrockClient.getFoundationModel(
                    r -> r.modelIdentifier(modelIdentifier)
            );

            FoundationModelDetails model = response.modelDetails();

            System.out.println(" Model ID:                     " + model.modelId());
            System.out.println(" Model ARN:                    " + model.modelArn());
            System.out.println(" Model Name:                   " + model.modelName());
            System.out.println(" Provider Name:                " + model.providerName());
            System.out.println(" Lifecycle status:             " + model.modelLifecycle().statusAsString());
            System.out.println(" Input modalities:             " + model.inputModalities());
            System.out.println(" Output modalities:            " + model.outputModalities());
            System.out.println(" Supported customizations:     " + model.customizationsSupported());
            System.out.println(" Supported inference types:    " + model.inferenceTypesSupported());
            System.out.println(" Response streaming supported: " + model.responseStreamingSupported());

            return model;

        } catch (ValidationException e) {
            throw new IllegalArgumentException(e.getMessage());
        } catch (SdkException e) {
            System.err.println(e.getMessage());
            throw new RuntimeException(e);
        }
    }

Get details about a foundation model using the asynchronous Amazon Bedrock client.

    /**
     * Get details about an Amazon Bedrock foundation model.
     *
     * @param bedrockClient   The async service client for accessing Amazon Bedrock.
     * @param modelIdentifier The model identifier.
     * @return An object containing the foundation model's details.
     */
    public static FoundationModelDetails getFoundationModel(BedrockAsyncClient bedrockClient, String modelIdentifier) {
        try {
            CompletableFuture<GetFoundationModelResponse> future = bedrockClient.getFoundationModel(
                    r -> r.modelIdentifier(modelIdentifier)
            );

            FoundationModelDetails model = future.get().modelDetails();

            System.out.println(" Model ID:                     " + model.modelId());
            System.out.println(" Model ARN:                    " + model.modelArn());
            System.out.println(" Model Name:                   " + model.modelName());
            System.out.println(" Provider Name:                " + model.providerName());
            System.out.println(" Lifecycle status:             " + model.modelLifecycle().statusAsString());
            System.out.println(" Input modalities:             " + model.inputModalities());
            System.out.println(" Output modalities:            " + model.outputModalities());
            System.out.println(" Supported customizations:     " + model.customizationsSupported());
            System.out.println(" Supported inference types:    " + model.inferenceTypesSupported());
            System.out.println(" Response streaming supported: " + model.responseStreamingSupported());

            return model;

        } catch (ExecutionException e) {
            if (e.getMessage().contains("ValidationException")) {
                throw new IllegalArgumentException(e.getMessage());
            } else {
                System.err.println(e.getMessage());
                throw new RuntimeException(e);
            }
        } catch (InterruptedException e) {
            Thread.currentThread().interrupt();
            System.err.println(e.getMessage());
            throw new RuntimeException(e);
        }
    }

• For API details, see

GetFoundationModel in _AWS SDK for Java 2.x API Reference_.

The following code example shows how to use ListFoundationModels.

SDK for Java 2.x

Note

There's more on GitHub. Find the complete example and learn how to set up and run in the AWS Code\ Examples Repository.

List the available Amazon Bedrock foundation models using the synchronous Amazon Bedrock client.

    /**
     * Lists Amazon Bedrock foundation models that you can use.
     * You can filter the results with the request parameters.
     *
     * @param bedrockClient The service client for accessing Amazon Bedrock.
     * @return A list of objects containing the foundation models' details
     */
    public static List<FoundationModelSummary> listFoundationModels(BedrockClient bedrockClient) {

        try {
            ListFoundationModelsResponse response = bedrockClient.listFoundationModels(r -> {});

            List<FoundationModelSummary> models = response.modelSummaries();

            if (models.isEmpty()) {
                System.out.println("No available foundation models in " + region.toString());
            } else {
                for (FoundationModelSummary model : models) {
                    System.out.println("Model ID: " + model.modelId());
                    System.out.println("Provider: " + model.providerName());
                    System.out.println("Name:     " + model.modelName());
                    System.out.println();
                }
            }

            return models;

        } catch (SdkClientException e) {
            System.err.println(e.getMessage());
            throw new RuntimeException(e);
        }
    }

List the available Amazon Bedrock foundation models using the asynchronous Amazon Bedrock client.

    /**
     * Lists Amazon Bedrock foundation models that you can use.
     * You can filter the results with the request parameters.
     *
     * @param bedrockClient The async service client for accessing Amazon Bedrock.
     * @return A list of objects containing the foundation models' details
     */
    public static List<FoundationModelSummary> listFoundationModels(BedrockAsyncClient bedrockClient) {
        try {
            CompletableFuture<ListFoundationModelsResponse> future = bedrockClient.listFoundationModels(r -> {});

            List<FoundationModelSummary> models = future.get().modelSummaries();

            if (models.isEmpty()) {
                System.out.println("No available foundation models in " + region.toString());
            } else {
                for (FoundationModelSummary model : models) {
                    System.out.println("Model ID: " + model.modelId());
                    System.out.println("Provider: " + model.providerName());
                    System.out.println("Name:     " + model.modelName());
                    System.out.println();
                }
            }

            return models;

        } catch (InterruptedException e) {
            Thread.currentThread().interrupt();
            System.err.println(e.getMessage());
            throw new RuntimeException(e);
        } catch (ExecutionException e) {
            System.err.println(e.getMessage());
            throw new RuntimeException(e);
        }
    }

• For API details, see

ListFoundationModels in _AWS SDK for Java 2.x API Reference_.

Document Conventions

AWS Batch

Amazon Bedrock Runtime

Did this page help you? - Yes

Thanks for letting us know we're doing a good job!

If you've got a moment, please tell us what we did right so we can do more of it.

Did this page help you? - No

Thanks for letting us know this page needs work. We're sorry we let you down.

If you've got a moment, please tell us how we can make the documentation better.

Supported foundation models in Amazon Bedrock - Amazon Bedrock

Supported foundation models in Amazon Bedrock - Amazon Bedrock

[Open PDF](https://docs.aws.amazon.com/bedrock/latest/userguide/models-supported.html/pdfs/bedrock/latest/userguide/bedrock-ug.pdf#models-supported "Open PDF")

Documentation Amazon Bedrock User Guide

Supported foundation models in Amazon Bedrock

Amazon Bedrock supports foundation models (FMs) from multiple providers.

The following table lists each model alongside the ID that you can use to make on-demand API calls, the AWS Regions that support it, its capabilities, and links to relevant documentation:

ProviderModel nameModel IDRegions supportedInput modalitiesOutput modalitiesStreaming supportedInference parametersHyperparametersAI21 LabsJamba 1.5 Largeai21.jamba-1-5-large-v1:0

us-east-1

TextText, ChatYesLinkN/AAI21 LabsJamba 1.5 Miniai21.jamba-1-5-mini-v1:0

us-east-1

TextText, ChatYesLinkN/AAmazonNova Canvasamazon.nova-canvas-v1:0

us-east-1

ap-northeast-1

eu-west-1

Text, ImageImageNoLinkLinkAmazonNova Liteamazon.nova-lite-v1:0

us-east-1

us-east-2\*

us-west-1\*

us-west-2\*

us-gov-west-1

ap-east-2\*

ap-northeast-1

ap-northeast-2\*

ap-south-1\*

ap-southeast-1\*

ap-southeast-2

ap-southeast-3

ap-southeast-4\*

ap-southeast-5\*

ap-southeast-7\*

ca-central-1\*

eu-central-1\*

eu-north-1

eu-south-1\*

eu-south-2\*

eu-west-1\*

eu-west-2

eu-west-3\*

il-central-1\*

me-central-1

Text, Image, VideoTextYesLinkLinkAmazonNova Microamazon.nova-micro-v1:0

us-east-1

us-east-2\*

us-west-2\*

us-gov-west-1

ap-east-2\*

ap-northeast-1\*

ap-northeast-2\*

ap-south-1\*

ap-southeast-1\*

ap-southeast-2

ap-southeast-3\*

ap-southeast-5\*

ap-southeast-7\*

eu-central-1\*

eu-north-1\*

eu-south-1\*

eu-south-2\*

eu-west-1\*

eu-west-2

eu-west-3\*

il-central-1\*

me-central-1\*

TextTextYesLinkLinkAmazonNova Premieramazon.nova-premier-v1:0

us-east-1\*

us-east-2\*

us-west-2\*

Text, Image, VideoTextYesLinkLinkAmazonNova Proamazon.nova-pro-v1:0

us-east-1

us-east-2\*

us-west-1\*

us-west-2\*

us-gov-west-1

ap-east-2\*

ap-northeast-1\*

ap-northeast-2\*

ap-south-1\*

ap-southeast-1\*

ap-southeast-2

ap-southeast-3

ap-southeast-4\*

ap-southeast-5\*

ap-southeast-7\*

eu-central-1\*

eu-north-1\*

eu-south-1\*

eu-south-2\*

eu-west-1\*

eu-west-2

eu-west-3\*

il-central-1\*

me-central-1

Text, Image, VideoTextYesLinkLinkAmazonNova Reelamazon.nova-reel-v1:0

us-east-1

ap-northeast-1

eu-west-1

Text, ImageVideoNoLinkLinkAmazonNova Reelamazon.nova-reel-v1:1

us-east-1

Text, ImageVideoNoLinkLinkAmazonNova Sonicamazon.nova-sonic-v1:0

us-east-1

ap-northeast-1

eu-north-1

SpeechSpeech, TextYesLinkLinkAmazonRerank 1.0amazon.rerank-v1:0

us-west-2

ap-northeast-1

ca-central-1

eu-central-1

TextTextNoN/AN/AAmazonTitan Embeddings G1 - Textamazon.titan-embed-text-v1

us-east-1

us-west-2

ap-northeast-1

eu-central-1

TextEmbeddingNoLinkN/AAmazonTitan Image Generator G1 v2amazon.titan-image-generator-v2:0

us-east-1

us-west-2

Text, ImageImageNoLinkLinkAmazonTitan Image Generator G1amazon.titan-image-generator-v1

us-east-1

us-west-2

ap-south-1

eu-west-1

eu-west-2

Text, ImageImageNoLinkLinkAmazonTitan Multimodal Embeddings G1amazon.titan-embed-image-v1

us-east-1

us-west-2

ap-south-1

ap-southeast-2

ca-central-1

eu-central-1

eu-west-1

eu-west-2

eu-west-3

sa-east-1

Text, ImageEmbeddingNoLinkLinkAmazonTitan Text Embeddings V2amazon.titan-embed-text-v2:0

us-east-1

us-east-2

us-west-2

us-gov-east-1

us-gov-west-1

ap-northeast-1

ap-northeast-2

ap-northeast-3

ap-south-1

ap-south-2

ap-southeast-2

ca-central-1

eu-central-1

eu-central-2

eu-north-1

eu-south-1

eu-south-2

eu-west-1

eu-west-2

eu-west-3

sa-east-1

TextEmbeddingNoLinkN/AAmazonTitan Text G1 - Expressamazon.titan-text-express-v1

us-east-1

us-west-2

us-gov-west-1

ap-northeast-1

ap-south-1

ap-southeast-2

ca-central-1

eu-central-1

eu-west-1

eu-west-2

eu-west-3

sa-east-1

TextText, ChatYesLinkLinkAmazonTitan Text G1 - Liteamazon.titan-text-lite-v1

us-east-1

us-west-2

ap-south-1

ap-southeast-2

ca-central-1

eu-central-1

eu-west-1

eu-west-2

eu-west-3

sa-east-1

TextTextYesLinkLinkAnthropicClaude 3 Haikuanthropic.claude-3-haiku-20240307-v1:0

us-east-1

us-east-2\*

us-west-2

us-gov-east-1\*

us-gov-west-1

ap-northeast-1

ap-northeast-2

ap-south-1

ap-southeast-1

ap-southeast-1

ap-southeast-2

ca-central-1

eu-central-1

eu-central-2

eu-west-1

eu-west-2

eu-west-3

sa-east-1

Text, ImageText, ChatYesLinkN/AAnthropicClaude 3 Opusanthropic.claude-3-opus-20240229-v1:0

us-east-1\*

us-west-2

Text, ImageText, ChatYesLinkN/AAnthropicClaude 3 Sonnetanthropic.claude-3-sonnet-20240229-v1:0

us-east-1

us-west-2

ap-northeast-1\*

ap-northeast-2\*

ap-south-1

ap-southeast-1\*

ap-southeast-2

ca-central-1

eu-central-1

eu-west-1

eu-west-2

eu-west-3

sa-east-1

Text, ImageText, ChatYesLinkN/AAnthropicClaude 3.5 Haikuanthropic.claude-3-5-haiku-20241022-v1:0

us-east-1\*

us-east-2\*

us-west-2

TextText, ChatYesLinkN/AAnthropicClaude 3.5 Sonnet v2anthropic.claude-3-5-sonnet-20241022-v2:0

us-east-1\*

us-east-2\*

us-west-2

ap-northeast-1\*

ap-northeast-2\*

ap-northeast-3\*

ap-south-1\*

ap-south-2\*

ap-southeast-1\*

ap-southeast-2

Text, ImageText, ChatYesLinkN/AAnthropicClaude 3.5 Sonnetanthropic.claude-3-5-sonnet-20240620-v1:0

us-east-1

us-east-2\*

us-west-2

us-gov-east-1\*

us-gov-west-1

ap-northeast-1

ap-northeast-2

ap-south-1\*

ap-southeast-1

ap-southeast-1

ap-southeast-2\*

eu-central-1

eu-central-2

eu-west-1\*

eu-west-3\*

Text, ImageText, ChatYesLinkN/AAnthropicClaude 3.7 Sonnetanthropic.claude-3-7-sonnet-20250219-v1:0

us-east-1\*

us-east-2\*

us-west-2\*

us-gov-east-1\*

us-gov-west-1

ap-northeast-1\*

ap-northeast-2\*

ap-northeast-3\*

ap-south-1\*

ap-south-2\*

ap-southeast-1\*

ap-southeast-2\*

eu-central-1\*

eu-north-1\*

eu-west-1\*

eu-west-2

eu-west-3\*

Text, ImageText, ChatYesLinkN/AAnthropicClaude Haiku 4.5anthropic.claude-haiku-4-5-20251001-v1:0

us-east-1\*

us-east-2\*

us-west-1\*

us-west-2\*

ap-northeast-1\*

ap-northeast-2\*

ap-northeast-3\*

ap-south-1\*

ap-south-2\*

ap-southeast-1\*

ap-southeast-2\*

ap-southeast-3\*

ap-southeast-4\*

ca-central-1\*

eu-central-1\*

eu-central-2\*

eu-north-1\*

eu-south-1\*

eu-south-2\*

eu-west-1\*

eu-west-2\*

eu-west-3\*

sa-east-1\*

Text, ImageText, ChatYesLinkN/AAnthropicClaude Opus 4.1anthropic.claude-opus-4-1-20250805-v1:0

us-east-1\*

us-east-2\*

us-west-2\*

Text, ImageText, ChatYesLinkN/AAnthropicClaude Opus 4anthropic.claude-opus-4-20250514-v1:0

us-east-1\*

us-east-2\*

us-west-2\*

Text, ImageText, ChatYesLinkN/AAnthropicClaude Sonnet 4.5anthropic.claude-sonnet-4-5-20250929-v1:0

us-east-1\*

us-east-2\*

us-west-1\*

us-west-2\*

ap-northeast-1\*

ap-northeast-2\*

ap-northeast-3\*

ap-south-1\*

ap-south-2\*

ap-southeast-1\*

ap-southeast-2\*

ap-southeast-3\*

ap-southeast-4\*

ca-central-1\*

eu-central-1\*

eu-central-2\*

eu-north-1\*

eu-south-1\*

eu-south-2\*

eu-west-1\*

eu-west-2\*

eu-west-3\*

sa-east-1\*

Text, ImageText, ChatYesLinkN/AAnthropicClaude Sonnet 4anthropic.claude-sonnet-4-20250514-v1:0

us-east-1\*

us-east-2\*

us-west-1\*

us-west-2\*

ap-east-2\*

ap-northeast-1\*

ap-northeast-2\*

ap-northeast-3\*

ap-south-1\*

ap-south-2\*

ap-southeast-1\*

ap-southeast-2\*

ap-southeast-3\*

ap-southeast-4\*

ap-southeast-5\*

ap-southeast-7\*

eu-central-1\*

eu-north-1\*

eu-south-1\*

eu-south-2\*

eu-west-1\*

eu-west-3\*

il-central-1\*

me-central-1\*

Text, ImageText, ChatYesLinkN/ACohereCommand R+cohere.command-r-plus-v1:0

us-east-1

us-west-2

TextText, ChatYesLinkN/ACohereCommand Rcohere.command-r-v1:0

us-east-1

us-west-2

TextText, ChatYesLinkN/ACohereEmbed Englishcohere.embed-english-v3

us-east-1

us-west-2

ap-northeast-1

ap-south-1

ap-southeast-1

ap-southeast-2

ca-central-1

eu-central-1

eu-west-1

eu-west-2

eu-west-3

sa-east-1

TextEmbeddingNoLinkN/ACohereEmbed Multilingualcohere.embed-multilingual-v3

us-east-1

us-west-2

ap-northeast-1

ap-south-1

ap-southeast-1

ap-southeast-2

ca-central-1

eu-central-1

eu-west-1

eu-west-2

eu-west-3

sa-east-1

TextEmbeddingNoLinkN/ACohereEmbed v4cohere.embed-v4:0

us-east-1

us-east-2\*

us-west-1\*

us-west-2\*

ap-northeast-1

ap-northeast-2\*

ap-northeast-3\*

ap-south-1\*

ap-south-2\*

ap-southeast-1\*

ap-southeast-2\*

ap-southeast-3\*

ap-southeast-4\*

ca-central-1\*

eu-central-1\*

eu-central-2\*

eu-north-1\*

eu-south-1\*

eu-south-2\*

eu-west-1

eu-west-2\*

eu-west-3\*

sa-east-1\*

Text, ImageEmbeddingNoLinkN/ACohereRerank 3.5cohere.rerank-v3-5:0

us-east-1

us-west-2

ap-northeast-1

ca-central-1

eu-central-1

TextTextNoLinkN/ADeepSeekDeepSeek-R1deepseek.r1-v1:0

us-east-1\*

us-east-2\*

us-west-2\*

TextTextYesLinkN/ADeepSeekDeepSeek-V3.1deepseek.v3-v1:0

us-east-2

us-west-2

ap-northeast-1

ap-south-1

ap-southeast-3

eu-north-1

eu-west-2

TextTextYesLinkN/ALuma AIRay v2luma.ray-v2:0

us-west-2

TextVideoNoLinkN/AMetaLlama 3 8B Instructmeta.llama3-8b-instruct-v1:0

us-east-1

us-west-2

us-gov-west-1

ap-south-1

ca-central-1

eu-west-2

TextText, ChatYesLinkN/AMetaLlama 3 70B Instructmeta.llama3-70b-instruct-v1:0

us-east-1

us-west-2

us-gov-west-1

ap-south-1

ca-central-1

eu-west-2

TextText, ChatYesLinkN/AMetaLlama 3.1 8B Instructmeta.llama3-1-8b-instruct-v1:0

us-east-1\*

us-east-2\*

us-west-2

TextText, ChatYesLinkN/AMetaLlama 3.1 70B Instructmeta.llama3-1-70b-instruct-v1:0

us-east-1\*

us-east-2\*

us-west-2

TextText, ChatYesLinkN/AMetaLlama 3.1 405B Instructmeta.llama3-1-405b-instruct-v1:0

us-east-2\*

us-west-2

TextText, ChatYesLinkN/AMetaLlama 3.2 1B Instructmeta.llama3-2-1b-instruct-v1:0

us-east-1\*

us-east-2\*

us-west-2\*

eu-central-1\*

eu-west-1\*

eu-west-3\*

TextText, ChatYesLinkN/AMetaLlama 3.2 3B Instructmeta.llama3-2-3b-instruct-v1:0

us-east-1\*

us-east-2\*

us-west-2\*

eu-central-1\*

eu-west-1\*

eu-west-3\*

TextText, ChatYesLinkN/AMetaLlama 3.2 11B Instructmeta.llama3-2-11b-instruct-v1:0

us-east-1\*

us-east-2\*

us-west-2\*

Text, ImageText, ChatYesLinkN/AMetaLlama 3.2 90B Instructmeta.llama3-2-90b-instruct-v1:0

us-east-1\*

us-east-2\*

us-west-2\*

Text, ImageText, ChatYesLinkN/AMetaLlama 3.3 70B Instructmeta.llama3-3-70b-instruct-v1:0

us-east-1\*

us-east-2

us-west-2\*

TextText, ChatYesLinkN/AMetaLlama 4 Maverick 17B Instructmeta.llama4-maverick-17b-instruct-v1:0

us-east-1\*

us-east-2\*

us-west-1\*

us-west-2\*

Text, ImageText, ChatYesLinkN/AMetaLlama 4 Scout 17B Instructmeta.llama4-scout-17b-instruct-v1:0

us-east-1\*

us-east-2\*

us-west-1\*

us-west-2\*

Text, ImageText, ChatYesLinkN/AMistral AIMistral 7B Instructmistral.mistral-7b-instruct-v0:2

us-east-1

us-west-2

ap-south-1

ap-southeast-2

ca-central-1

eu-west-1

eu-west-2

eu-west-3

sa-east-1

TextTextYesLinkN/AMistral AIMistral Large (24.02)mistral.mistral-large-2402-v1:0

us-east-1

us-west-2

ap-south-1

ap-southeast-2

ca-central-1

eu-west-1

eu-west-2

eu-west-3

sa-east-1

TextTextYesLinkN/AMistral AIMistral Large (24.07)mistral.mistral-large-2407-v1:0

us-west-2

TextTextYesLinkN/AMistral AIMistral Small (24.02)mistral.mistral-small-2402-v1:0

us-east-1

TextTextYesLinkN/AMistral AIMixtral 8x7B Instructmistral.mixtral-8x7b-instruct-v0:1

us-east-1

us-west-2

ap-south-1

ap-southeast-2

ca-central-1

eu-west-1

eu-west-2

eu-west-3

sa-east-1

TextTextYesLinkN/AMistral AIPixtral Large (25.02)mistral.pixtral-large-2502-v1:0

us-east-1\*

us-east-2\*

us-west-2\*

eu-central-1\*

eu-north-1\*

eu-west-1\*

eu-west-3\*

Text, ImageTextYesLinkN/AOpenAIgpt-oss-120bopenai.gpt-oss-120b-1:0

us-east-1

us-east-2

us-west-2

ap-northeast-1

ap-south-1

ap-southeast-3

eu-central-1

eu-north-1

eu-south-1

eu-west-1

eu-west-2

sa-east-1

TextTextYesLinkN/AOpenAIgpt-oss-20bopenai.gpt-oss-20b-1:0

us-east-1

us-east-2

us-west-2

ap-northeast-1

ap-south-1

ap-southeast-3

eu-central-1

eu-north-1

eu-south-1

eu-west-1

eu-west-2

sa-east-1

TextTextYesLinkN/AQwenQwen3 235B A22B 2507qwen.qwen3-235b-a22b-2507-v1:0

us-east-2

us-west-2

ap-northeast-1

ap-south-1

ap-southeast-3

eu-central-1

eu-north-1

eu-south-1

eu-west-2

TextTextYesLinkN/AQwenQwen3 32B (dense)qwen.qwen3-32b-v1:0

us-east-1

us-east-2

us-west-2

ap-northeast-1

ap-south-1

ap-southeast-3

eu-central-1

eu-north-1

eu-south-1

eu-west-1

eu-west-2

sa-east-1

TextTextYesLinkN/AQwenQwen3 Coder 480B A35B Instructqwen.qwen3-coder-480b-a35b-v1:0

us-east-2

us-west-2

ap-northeast-1

ap-south-1

ap-southeast-3

eu-north-1

eu-west-2

TextTextYesLinkN/AQwenQwen3-Coder-30B-A3B-Instructqwen.qwen3-coder-30b-a3b-v1:0

us-east-1

us-east-2

us-west-2

ap-northeast-1

ap-south-1

ap-southeast-3

eu-central-1

eu-north-1

eu-south-1

eu-west-1

eu-west-2

sa-east-1

TextTextYesLinkN/AStability AIStable Diffusion 3.5 Largestability.sd3-5-large-v1:0

us-west-2

Text, ImageImageNoLinkN/AStability AIStable Image Control Sketchstability.stable-image-control-sketch-v1:0

us-east-1\*

us-east-2\*

us-west-2\*

Text, ImageImageNoLinkN/AStability AIStable Image Control Structurestability.stable-image-control-structure-v1:0

us-east-1\*

us-east-2\*

us-west-2\*

Text, ImageImageNoLinkN/AStability AIStable Image Core 1.0stability.stable-image-core-v1:1

us-west-2

TextImageNoLinkN/AStability AIStable Image Erase Objectstability.stable-image-erase-object-v1:0

us-east-1\*

us-east-2\*

us-west-2\*

Text, ImageImageNoLinkN/AStability AIStable Image Inpaintstability.stable-image-inpaint-v1:0

us-east-1\*

us-east-2\*

us-west-2\*

Text, ImageImageNoLinkN/AStability AIStable Image Remove Backgroundstability.stable-image-remove-background-v1:0

us-east-1\*

us-east-2\*

us-west-2\*

Text, ImageImageNoLinkN/AStability AIStable Image Search and Recolorstability.stable-image-search-recolor-v1:0

us-east-1\*

us-east-2\*

us-west-2\*

Text, ImageImageNoLinkN/AStability AIStable Image Search and Replacestability.stable-image-search-replace-v1:0

us-east-1\*

us-east-2\*

us-west-2\*

Text, ImageImageNoLinkN/AStability AIStable Image Style Guidestability.stable-image-style-guide-v1:0

us-east-1\*

us-east-2\*

us-west-2\*

Text, ImageImageNoLinkN/AStability AIStable Image Style Transferstability.stable-style-transfer-v1:0

us-east-1\*

us-east-2\*

us-west-2\*

Text, ImageImageNoLinkN/AStability AIStable Image Ultra 1.0stability.stable-image-ultra-v1:1

us-west-2

TextImageNoLinkN/ATwelveLabsMarengo Embed v2.7twelvelabs.marengo-embed-2-7-v1:0

us-east-1\*

ap-northeast-2\*

eu-west-1\*

Text, Image, Speech, VideoEmbeddingNoLinkN/ATwelveLabsPegasus v1.2twelvelabs.pegasus-1-2-v1:0

us-east-1\*

us-west-2\*

ap-northeast-2\*

eu-west-1\*

Text, VideoTextYesLinkN/AWriterPalmyra X4writer.palmyra-x4-v1:0

us-west-2\*

TextTextYesLinkN/AWriterPalmyra X5writer.palmyra-x5-v1:0

us-west-2\*

TextTextYesLinkN/A

Note

\\* Some models are accessible in some Regions only through cross-Region inference. To learn more about cross-Region inference, see Increase throughput with cross-Region inference and Supported Regions and models for inference profiles.

Certain models have a targetted date for deprecation. For more information, see Model lifecycle:

Document Conventions

Get model information

Model support by Region

Did this page help you? - Yes

Thanks for letting us know we're doing a good job!

If you've got a moment, please tell us what we did right so we can do more of it.

Did this page help you? - No

Thanks for letting us know this page needs work. We're sorry we let you down.

If you've got a moment, please tell us how we can make the documentation better.

Model Reference

Supported Foundation Models

Amazon Models

Amazon Titan Text

Model ID: amazon.titan-text-express-v1

• Description: High-quality text generation model
• Context Window: Up to 8K tokens
• Languages: English, Spanish, French, German, Italian, Portuguese

Payload Format:

{
    "inputText": "Your prompt here",
    "textGenerationConfig": {
        "maxTokenCount": 512,
        "temperature": 0.7,
        "topP": 0.9
    }
}

Response Format:

{
    "results": [{
        "outputText": "Generated text"
    }]
}

Amazon Titan Text Lite

Model ID: amazon.titan-text-lite-v1

• Description: Cost-effective text generation model
• Context Window: Up to 4K tokens
• Use Case: Simple text generation tasks

Amazon Titan Embeddings

Model ID: amazon.titan-embed-text-v1

• Description: High-quality text embeddings
• Context Window: 8K tokens
• Output: 1024-dimensional vector

Payload Format:

{
    "inputText": "Your text here"
}

Response Format:

{
    "embedding": [0.1, -0.2, 0.3, ...]
}

Amazon Titan Image Generator

Model ID: amazon.titan-image-generator-v1

• Description: High-quality image generation
• Image Size: 512x512, 1024x1024
• Use Case: Text-to-image generation

Payload Format:

{
    "taskType": "TEXT_IMAGE",
    "textToImageParams": {
        "text": "Your description"
    },
    "imageGenerationConfig": {
        "numberOfImages": 1,
        "quality": "standard",
        "cfgScale": 8.0,
        "height": 512,
        "width": 512,
        "seed": 12345
    }
}

Anthropic Models

Claude 3.5 Sonnet

Model ID: anthropic.claude-3-5-sonnet-20241022-v2:0

• Description: High-performance model for complex reasoning, analysis, and creative tasks
• Context Window: 200K tokens
• Languages: Multiple languages supported
• Use Case: Code generation, complex analysis, creative writing, research
• Features: Tool use, function calling, JSON mode

Payload Format:

{
    "anthropic_version": "bedrock-2023-05-31",
    "max_tokens": 1000,
    "messages": [{
        "role": "user",
        "content": "Your message"
    }]
}

Response Format:

{
    "content": [{
        "text": "Response content"
    }],
    "usage": {
        "input_tokens": 10,
        "output_tokens": 20
    }
}

Claude 3.5 Haiku

Model ID: anthropic.claude-3-5-haiku-20241022-v2:0

• Description: Fast and affordable model for real-time applications
• Context Window: 200K tokens
• Use Case: Real-time applications, chatbots, quick responses
• Features: Tool use, function calling, JSON mode

Claude 3 Opus

Model ID: anthropic.claude-3-opus-20240229-v1:0

• Description: Most capable model
• Context Window: 200K tokens
• Use Case: Complex reasoning, analysis

Claude 3 Sonnet (Legacy)

Model ID: anthropic.claude-3-sonnet-20240229-v1:0

• Description: Previous generation model
• Context Window: 200K tokens
• Use Case: General purpose applications

Meta Models

Llama 3.1 70B

Model ID: meta.llama3-1-70b-instruct-v1:0

• Description: Latest generation large open-source model
• Context Window: 128K tokens
• Use Case: General purpose instruction following, complex reasoning
• Features: Improved instruction following, larger context window

Llama 3.1 8B

Model ID: meta.llama3-1-8b-instruct-v1:0

• Description: Latest generation small fast model
• Context Window: 8K tokens
• Use Case: Fast inference, lightweight applications

Llama 3 70B

Model ID: meta.llama3-70b-instruct-v1:0

• Description: Previous generation large open-source model
• Context Window: 8K tokens
• Use Case: General purpose instruction following

Payload Format:

{
    "prompt": "[INST] Your prompt here [/INST]",
    "max_gen_len": 512,
    "temperature": 0.7,
    "top_p": 0.9
}

Response Format:

{
    "generation": "Generated text"
}

Llama 3 8B

Model ID: meta.llama3-8b-instruct-v1:0

• Description: Smaller, faster version
• Context Window: 8K tokens
• Use Case: Fast inference, lightweight applications

Stability AI Models

Stable Diffusion XL

Model ID: stability.stable-diffusion-xl-v1

• Description: High-quality image generation
• Image Size: Up to 1024x1024
• Use Case: Text-to-image generation, art creation

Payload Format:

{
    "text_prompts": [{
        "text": "Your description"
    }],
    "style_preset": "photographic",
    "seed": 12345,
    "cfg_scale": 10,
    "steps": 50
}

Response Format:

{
    "artifacts": [{
        "base64": "base64-encoded-image-data",
        "finishReason": "SUCCESS"
    }]
}

Other Models

Cohere Command

Model ID: cohere.command-text-v14

• Description: Text generation model
• Context Window: 128K tokens
• Use Case: Content generation, summarization

Mistral Models

Model ID: mistral.mistral-7b-instruct-v0:2

• Description: High-performing open-source model
• Context Window: 32K tokens
• Use Case: Instruction following, code generation

Model ID: mistral.mixtral-8x7b-instruct-v0:1

• Description: Mixture of experts model
• Context Window: 32K tokens
• Use Case: Complex reasoning tasks

Model Selection Guide

Use Case Recommendations

Performance Characteristics

Model Comparison Matrix

Common Model ID Constants

// Claude Models
public static final String CLAUDE_SONNET_45 = "anthropic.claude-sonnet-4-5-20250929-v1:0";
public static final String CLAUDE_HAIKU_45 = "anthropic.claude-haiku-4-5-20251001-v1:0";
public static final String CLAUDE_OPUS_41 = "anthropic.claude-opus-4-1-20250805-v1:0";
public static final String CLAUDE_37_SONNET = "anthropic.claude-3-7-sonnet-20250219-v1:0";
public static final String CLAUDE_35_SONNET_V2 = "anthropic.claude-3-5-sonnet-20241022-v2:0";
public static final String CLAUDE_35_HAIKU = "anthropic.claude-3-5-haiku-20241022-v1:0";

// Llama Models
public static final String LLAMA_33_70B = "meta.llama3-3-70b-instruct-v1:0";
public static final String LLAMA_32_11B = "meta.llama3-2-11b-instruct-v1:0";
public static final String LLAMA_31_405B = "meta.llama3-1-405b-instruct-v1:0";
public static final String LLAMA_31_70B = "meta.llama3-1-70b-instruct-v1:0";
public static final String LLAMA_31_8B = "meta.llama3-1-8b-instruct-v1:0";

// Amazon Titan Models
public static final String TITAN_TEXT_EXPRESS = "amazon.titan-text-express-v1";
public static final String TITAN_TEXT_LITE = "amazon.titan-text-lite-v1";
public static final String TITAN_EMBEDDINGS = "amazon.titan-embed-text-v1";
public static final String TITAN_IMAGE_GENERATOR = "amazon.titan-image-generator-v1";

// Amazon Nova Models
public static final String NOVA_PREMIER = "amazon.nova-premier-v1:0";
public static final String NOVA_PRO = "amazon.nova-pro-v1:0";
public static final String NOVA_LITE = "amazon.nova-lite-v1:0";
public static final String NOVA_CANVAS = "amazon.nova-canvas-v1:0";

// Other Models
public static final String STABLE_DIFFUSION_XL = "stability.stable-diffusion-xl-v1";
public static final String MISTRAL_LARGE_2407 = "mistral.mistral-large-2407-v1:0";
public static final String COHERE_COMMAND = "cohere.command-text-v14";

Model Configuration Templates

Text Generation Template

private static JSONObject createTextGenerationPayload(String modelId, String prompt) {
    JSONObject payload = new JSONObject();

    if (modelId.startsWith("anthropic.claude")) {
        payload.put("anthropic_version", "bedrock-2023-05-31");
        payload.put("max_tokens", 1000);
        payload.put("messages", new JSONObject[]{new JSONObject()
            .put("role", "user")
            .put("content", prompt)
        });
    } else if (modelId.startsWith("meta.llama")) {
        payload.put("prompt", "[INST] " + prompt + " [/INST]");
        payload.put("max_gen_len", 512);
    } else if (modelId.startsWith("amazon.titan")) {
        payload.put("inputText", prompt);
        payload.put("textGenerationConfig", new JSONObject()
            .put("maxTokenCount", 512)
            .put("temperature", 0.7)
        );
    }

    return payload;
}

Image Generation Template

private static JSONObject createImageGenerationPayload(String modelId, String prompt) {
    JSONObject payload = new JSONObject();

    if (modelId.equals("amazon.titan-image-generator-v1")) {
        payload.put("taskType", "TEXT_IMAGE");
        payload.put("textToImageParams", new JSONObject().put("text", prompt));
        payload.put("imageGenerationConfig", new JSONObject()
            .put("numberOfImages", 1)
            .put("quality", "standard")
            .put("height", 512)
            .put("width", 512)
        );
    } else if (modelId.equals("stability.stable-diffusion-xl-v1")) {
        payload.put("text_prompts", new JSONObject[]{new JSONObject().put("text", prompt)});
        payload.put("style_preset", "photographic");
        payload.put("steps", 50);
        payload.put("cfg_scale", 10);
    }

    return payload;
}

Model ID Lookup Guide

This document provides quick lookup for the most commonly used model IDs in Amazon Bedrock.

Text Generation Models

Claude (Anthropic)

Llama (Meta)

Mistral AI

Amazon

Cohere

Image Generation Models

Stability AI

Amazon Nova

Embedding Models

Amazon

Cohere

Selection Guide

By Speed

1. Fastest: Llama 3.2 1B, Claude 4.5 Haiku, Titan Lite 2. Fast: Mistral 7B, Llama 3.2 3B 3. Medium: Claude 3.5 Sonnet, Llama 3.2 11B 4. Slow: Claude 4.5 Sonnet, Llama 3.3 70B

By Quality

1. Highest: Claude 4.5 Sonnet, Claude 3.7 Sonnet, Claude Opus 4.1 2. High: Claude 3.5 Sonnet, Llama 3.3 70B 3. Medium: Mistral Large, Llama 3.2 11B 4. Basic: Mistral 7B, Llama 3.2 3B

By Cost

1. Most Affordable: Claude 4.5 Haiku, Llama 3.2 1B 2. Affordable: Mistral 7B, Titan Lite 3. Medium: Claude 3.5 Haiku, Llama 3.2 3B 4. Expensive: Claude 4.5 Sonnet, Llama 3.3 70B

Common Patterns

Default Model Selection

// For most applications
String DEFAULT_MODEL = "anthropic.claude-sonnet-4-5-20250929-v1:0";

// For real-time applications
String FAST_MODEL = "anthropic.claude-haiku-4-5-20251001-v1:0";

// For budget-sensitive applications
String CHEAP_MODEL = "amazon.titan-text-lite-v1";

// For complex reasoning
String POWERFUL_MODEL = "anthropic.claude-3-7-sonnet-20250219-v1:0";

Model Fallback Chain

private static final String[] MODEL_CHAIN = {
    "anthropic.claude-sonnet-4-5-20250929-v1:0",  // Primary
    "anthropic.claude-haiku-4-5-20251001-v1:0",  // Fast fallback
    "amazon.titan-text-lite-v1"                 // Cheap fallback
};

Testing Strategies

Unit Testing

Mocking Bedrock Clients

@ExtendWith(MockitoExtension.class)
class BedrockServiceTest {

    @Mock
    private BedrockRuntimeClient bedrockRuntimeClient;

    @InjectMocks
    private BedrockAIService aiService;

    @Test
    void shouldGenerateTextWithClaude() {
        // Arrange
        String modelId = "anthropic.claude-3-sonnet-20240229-v1:0";
        String prompt = "Hello, world!";
        String expectedResponse = "Hello! How can I help you today?";

        InvokeModelResponse mockResponse = InvokeModelResponse.builder()
            .body(SdkBytes.fromUtf8String(
                "{\"content\":[{\"text\":\"" + expectedResponse + "\"}]}"))
            .build();

        when(bedrockRuntimeClient.invokeModel(any(InvokeModelRequest.class)))
            .thenReturn(mockResponse);

        // Act
        String result = aiService.generateText(prompt, modelId);

        // Assert
        assertThat(result).isEqualTo(expectedResponse);
        verify(bedrockRuntimeClient).invokeModel(argThat(request ->
            request.modelId().equals(modelId)));
    }

    @Test
    void shouldHandleThrottling() {
        // Arrange
        when(bedrockRuntimeClient.invokeModel(any(InvokeModelRequest.class)))
            .thenThrow(ThrottlingException.builder()
                .message("Rate limit exceeded")
                .build());

        // Act & Assert
        assertThatThrownBy(() -> aiService.generateText("test"))
            .isInstanceOf(RuntimeException.class)
            .hasMessageContaining("Rate limit exceeded");
    }
}

Testing Error Conditions

@Test
void shouldHandleInvalidModelId() {
    String invalidModelId = "invalid.model.id";
    String prompt = "test";

    when(bedrockRuntimeClient.invokeModel(any(InvokeModelRequest.class)))
        .thenThrow(ValidationException.builder()
            .message("Invalid model identifier")
            .build());

    assertThatThrownBy(() -> aiService.generateText(prompt, invalidModelId))
        .isInstanceOf(IllegalArgumentException.class)
        .hasMessageContaining("Invalid model identifier");
}

Testing Multiple Models

@ParameterizedTest
@EnumSource(ModelProvider.class)
void shouldSupportAllModels(ModelProvider modelProvider) {
    String prompt = "Hello";
    String modelId = modelProvider.getModelId();
    String expectedResponse = "Response";

    InvokeModelResponse mockResponse = InvokeModelResponse.builder()
        .body(SdkBytes.fromUtf8String(createMockResponse(modelProvider, expectedResponse)))
        .build();

    when(bedrockRuntimeClient.invokeModel(any(InvokeModelRequest.class)))
        .thenReturn(mockResponse);

    String result = aiService.generateText(prompt, modelId);

    assertThat(result).isEqualTo(expectedResponse);
}

private enum ModelProvider {
    CLAUDE("anthropic.claude-3-sonnet-20240229-v1:0"),
    LLAMA("meta.llama3-70b-instruct-v1:0"),
    TITAN("amazon.titan-text-express-v1");

    private final String modelId;

    ModelProvider(String modelId) {
        this.modelId = modelId;
    }

    public String getModelId() {
        return modelId;
    }
}

Integration Testing

Testcontainers Integration

@Testcontainers
@SpringBootTest(classes = BedrockConfiguration.class)
@ActiveProfiles("test")
class BedrockIntegrationTest {

    @Container
    static LocalStackContainer localStack = new LocalStackContainer(
        DockerImageName.parse("localstack/localstack:latest"))
        .withServices(AWSService BEDROCK_RUNTIME)
        .withEnv("DEFAULT_REGION", "us-east-1");

    @Autowired
    private BedrockRuntimeClient bedrockRuntimeClient;

    @Test
    void shouldConnectToLocalStack() {
        assertThat(bedrockRuntimeClient).isNotNull();
    }

    @Test
    void shouldListFoundationModels() {
        ListFoundationModelsResponse response =
            bedrockRuntimeClient.listFoundationModels();

        assertThat(response.modelSummaries()).isNotEmpty();
    }
}

LocalStack Configuration

@Configuration
public class LocalStackConfig {

    @Value("${localstack.enabled:true}")
    private boolean localStackEnabled;

    @Bean
    @ConditionalOnProperty(name = "localstack.enabled", havingValue = "true")
    public AwsCredentialsProvider localStackCredentialsProvider() {
        return StaticCredentialsProvider.create(
            new AwsBasicCredentialsAccessKey("test", "test"));
    }

    @Bean
    @ConditionalOnProperty(name = "localstack.enabled", havingValue = "true")
    public BedrockRuntimeClient localStackBedrockRuntimeClient(
            AwsCredentialsProvider credentialsProvider) {

        return BedrockRuntimeClient.builder()
            .credentialsProvider(credentialsProvider)
            .endpointOverride(localStack.getEndpoint())
            .region(Region.US_EAST_1)
            .build();
    }
}

Performance Testing

@Test
void shouldPerformWithinTimeLimit() {
    String prompt = "Performance test prompt";
    int iterationCount = 100;

    long startTime = System.currentTimeMillis();

    for (int i = 0; i < iterationCount; i++) {
        InvokeModelResponse response = bedrockRuntimeClient.invokeModel(
            request -> request
                .modelId("anthropic.claude-3-sonnet-20240229-v1:0")
                .body(SdkBytes.fromUtf8String(createPayload(prompt))));
    }

    long duration = System.currentTimeMillis() - startTime;
    double avgTimePerRequest = (double) duration / iterationCount;

    assertThat(avgTimePerRequest).isLessThan(5000); // Less than 5 seconds per request
    System.out.println("Average response time: " + avgTimePerRequest + "ms");
}

Testing Streaming Responses

Streaming Handler Testing

@Test
void shouldStreamResponse() throws InterruptedException {
    String prompt = "Stream this response";

    MockStreamHandler mockHandler = new MockStreamHandler();

    InvokeModelWithResponseStreamRequest streamRequest =
        InvokeModelWithResponseStreamRequest.builder()
            .modelId("anthropic.claude-3-sonnet-20240229-v1:0")
            .body(SdkBytes.fromUtf8String(createPayload(prompt)))
            .build();

    bedrockRuntimeClient.invokeModelWithResponseStream(streamRequest, mockHandler);

    // Wait for streaming to complete
    mockHandler.awaitCompletion(10, TimeUnit.SECONDS);

    assertThat(mockHandler.getStreamedContent()).isNotEmpty();
    assertThat(mockHandler.getStreamedContent()).contains(" streamed");
}

private static class MockStreamHandler extends
    InvokeModelWithResponseStreamResponseHandler.Visitor {

    private final StringBuilder contentBuilder = new StringBuilder();
    private final CountDownLatch latch = new CountDownLatch(1);

    @Override
    public void visit(EventStream eventStream) {
        eventStream.forEach(event -> {
            if (event instanceof PayloadPart) {
                PayloadPart payloadPart = (PayloadPart) event;
                String chunk = payloadPart.bytes().asUtf8String();
                contentBuilder.append(chunk);
            }
        });
        latch.countDown();
    }

    public String getStreamedContent() {
        return contentBuilder.toString();
    }

    public void awaitCompletion(long timeout, TimeUnit unit)
        throws InterruptedException {
        latch.await(timeout, unit);
    }
}

Testing Configuration

Testing Different Regions

@ParameterizedTest
@EnumSource(value = Region.class,
    names = {"US_EAST_1", "US_WEST_2", "EU_WEST_1"})
void shouldWorkInAllRegions(Region region) {
    BedrockRuntimeClient client = BedrockRuntimeClient.builder()
        .region(region)
        .build();

    assertThat(client).isNotNull();
}

### Testing Authentication

@Test void shouldUseIamRoleForAuthentication() { BedrockRuntimeClient client = BedrockRuntimeClient.builder() .region(Region.US_EAST_1) .build();

// Test that client can make basic calls ListFoundationModelsResponse response = client.listFoundationModels();

assertThat(response).isNotNull(); }

## Test Data Management

### Test Response Fixtures

public class BedrockTestFixtures {

public static String createClaudeResponse() { return "{\"content\":[{\"text\":\"Hello! How can I help you today?\"}]}"; }

public static String createLlamaResponse() { return "{\"generation\":\"Hello! How can I assist you?\"}"; }

public static String createTitanResponse() { return "{\"results\":[{\"outputText\":\"Hello! How can I help?\"}]}"; }

public static String createPayload(String prompt) { return new JSONObject() .put("anthropic_version", "bedrock-2023-05-31") .put("max_tokens", 1000) .put("messages", new JSONObject[]{ new JSONObject() .put("role", "user") .put("content", prompt) }) .toString(); } }

### Integration Test Suite

@Suite @SelectClasses({ BedrockAIServiceTest.class, BedrockConfigurationTest.class, BedrockStreamingTest.class, BedrockErrorHandlingTest.class }) public class BedrockTestSuite { // Integration test suite for all Bedrock functionality }

## Testing Guidelines

### Unit Testing Best Practices

1. **Mock External Dependencies:** Always mock AWS SDK clients in unit tests
2. **Test Error Scenarios:** Include tests for throttling, validation errors, and network issues
3. **Parameterized Tests:** Test multiple models and configurations efficiently
4. **Performance Assertions:** Include basic performance benchmarks
5. **Test Data Fixtures:** Reuse test response data across tests

### Integration Testing Best Practices

1. **Use LocalStack:** Test against LocalStack for local development
2. **Test Multiple Regions:** Verify functionality across different AWS regions
3. **Test Edge Cases:** Include timeout, retry, and concurrent request scenarios
4. **Monitor Performance:** Track response times and error rates
5. **Clean Up Resources:** Ensure proper cleanup after integration tests

### Testing Configuration

application-test.properties

localstack.enabled=true aws.region=us-east-1 bedrock.timeout=5000 bedrock.retry.max-attempts=3