OpenAI Java API Library

Note

The OpenAI Java API Library is currently in beta.

There may be minor breaking changes.

Have thoughts or feedback? File an issue or comment on this thread.

The OpenAI Java SDK provides convenient access to the OpenAI REST API from applications written in Java.

The REST API documentation can be found on platform.openai.com. Javadocs are also available on javadoc.io.

Installation

Gradle

implementation("com.openai:openai-java:0.26.1")

Maven

<dependency>
    <groupId>com.openai</groupId>
    <artifactId>openai-java</artifactId>
    <version>0.26.1</version>
</dependency>

Requirements

This library requires Java 8 or later.

Usage

See the openai-java-example directory for complete and runnable examples.

import com.openai.client.OpenAIClient;
import com.openai.client.okhttp.OpenAIOkHttpClient;
import com.openai.models.ChatCompletion;
import com.openai.models.ChatCompletionCreateParams;
import com.openai.models.ChatModel;

// Configures using the `OPENAI_API_KEY`, `OPENAI_ORG_ID` and `OPENAI_PROJECT_ID` environment variables
OpenAIClient client = OpenAIOkHttpClient.fromEnv();

ChatCompletionCreateParams params = ChatCompletionCreateParams.builder()
    .addUserMessage("Say this is a test")
    .model(ChatModel.O3_MINI)
    .build();
ChatCompletion chatCompletion = client.chat().completions().create(params);

Client configuration

Configure the client using environment variables:

import com.openai.client.OpenAIClient;
import com.openai.client.okhttp.OpenAIOkHttpClient;

// Configures using the `OPENAI_API_KEY`, `OPENAI_ORG_ID` and `OPENAI_PROJECT_ID` environment variables
OpenAIClient client = OpenAIOkHttpClient.fromEnv();

Or manually:

import com.openai.client.OpenAIClient;
import com.openai.client.okhttp.OpenAIOkHttpClient;

OpenAIClient client = OpenAIOkHttpClient.builder()
    .apiKey("My API Key")
    .build();

Or using a combination of the two approaches:

import com.openai.client.OpenAIClient;
import com.openai.client.okhttp.OpenAIOkHttpClient;

OpenAIClient client = OpenAIOkHttpClient.builder()
    // Configures using the `OPENAI_API_KEY`, `OPENAI_ORG_ID` and `OPENAI_PROJECT_ID` environment variables
    .fromEnv()
    .apiKey("My API Key")
    .build();

See this table for the available options:

Setter	Environment variable	Required	Default value
apiKey	OPENAI_API_KEY	true	-
organization	OPENAI_ORG_ID	false	-
project	OPENAI_PROJECT_ID	false	-

Tip

Don't create more than one client in the same application. Each client has a connection pool and thread pools, which are more efficient to share between requests.

Requests and responses

To send a request to the OpenAI API, build an instance of some Params class and pass it to the corresponding client method. When the response is received, it will be deserialized into an instance of a Java class.

For example, client.chat().completions().create(...) should be called with an instance of ChatCompletionCreateParams, and it will return an instance of ChatCompletion.

Immutability

Each class in the SDK has an associated builder or factory method for constructing it.

Each class is immutable once constructed. If the class has an associated builder, then it has a toBuilder() method, which can be used to convert it back to a builder for making a modified copy.

Because each class is immutable, builder modification will never affect already built class instances.

Asynchronous execution

The default client is synchronous. To switch to asynchronous execution, call the async() method:

import com.openai.client.OpenAIClient;
import com.openai.client.okhttp.OpenAIOkHttpClient;
import com.openai.models.ChatCompletion;
import com.openai.models.ChatCompletionCreateParams;
import com.openai.models.ChatModel;
import java.util.concurrent.CompletableFuture;

// Configures using the `OPENAI_API_KEY`, `OPENAI_ORG_ID` and `OPENAI_PROJECT_ID` environment variables
OpenAIClient client = OpenAIOkHttpClient.fromEnv();

ChatCompletionCreateParams params = ChatCompletionCreateParams.builder()
    .addUserMessage("Say this is a test")
    .model(ChatModel.O3_MINI)
    .build();
CompletableFuture<ChatCompletion> chatCompletion = client.async().chat().completions().create(params);

Or create an asynchronous client from the beginning:

import com.openai.client.OpenAIClientAsync;
import com.openai.client.okhttp.OpenAIOkHttpClientAsync;
import com.openai.models.ChatCompletion;
import com.openai.models.ChatCompletionCreateParams;
import com.openai.models.ChatModel;
import java.util.concurrent.CompletableFuture;

// Configures using the `OPENAI_API_KEY`, `OPENAI_ORG_ID` and `OPENAI_PROJECT_ID` environment variables
OpenAIClientAsync client = OpenAIOkHttpClientAsync.fromEnv();

ChatCompletionCreateParams params = ChatCompletionCreateParams.builder()
    .addUserMessage("Say this is a test")
    .model(ChatModel.O3_MINI)
    .build();
CompletableFuture<ChatCompletion> chatCompletion = client.chat().completions().create(params);

The asynchronous client supports the same options as the synchronous one, except most methods return CompletableFutures.

Streaming

The SDK defines methods that return response "chunk" streams, where each chunk can be individually processed as soon as it arrives instead of waiting on the full response. Streaming methods generally correspond to SSE or JSONL responses.

Some of these methods may have streaming and non-streaming variants, but a streaming method will always have a Streaming suffix in its name, even if it doesn't have a non-streaming variant.

These streaming methods return StreamResponse for synchronous clients:

import com.openai.core.http.StreamResponse;
import com.openai.models.ChatCompletionChunk;

try (StreamResponse<ChatCompletionChunk> streamResponse = client.chat().completions().createStreaming(params)) {
    streamResponse.stream().forEach(chunk -> {
        System.out.println(chunk);
    });
    System.out.println("No more chunks!");
}

Or AsyncStreamResponse for asynchronous clients:

import com.openai.core.http.AsyncStreamResponse;
import com.openai.models.ChatCompletionChunk;
import java.util.Optional;

client.async().chat().completions().createStreaming(params).subscribe(chunk -> {
    System.out.println(chunk);
});

// If you need to handle errors or completion of the stream
client.async().chat().completions().createStreaming(params).subscribe(new AsyncStreamResponse.Handler<>() {
    @Override
    public void onNext(ChatCompletionChunk chunk) {
        System.out.println(chunk);
    }

    @Override
    public void onComplete(Optional<Throwable> error) {
        if (error.isPresent()) {
            System.out.println("Something went wrong!");
            throw new RuntimeException(error.get());
        } else {
            System.out.println("No more chunks!");
        }
    }
});

// Or use futures
client.async().chat().completions().createStreaming(params)
    .subscribe(chunk -> {
        System.out.println(chunk);
    })
    .onCompleteFuture();
    .whenComplete((unused, error) -> {
        if (error != null) {
            System.out.println("Something went wrong!");
            throw new RuntimeException(error);
        } else {
            System.out.println("No more chunks!");
        }
    });

Async streaming uses a dedicated per-client cached thread pool Executor to stream without blocking the current thread. This default is suitable for most purposes.

To use a different Executor, configure the subscription using the executor parameter:

import java.util.concurrent.Executor;
import java.util.concurrent.Executors;

Executor executor = Executors.newFixedThreadPool(4);
client.async().chat().completions().createStreaming(params).subscribe(
    chunk -> System.out.println(chunk), executor
);

Or configure the client globally using the streamHandlerExecutor method:

import com.openai.client.OpenAIClient;
import com.openai.client.okhttp.OpenAIOkHttpClient;
import java.util.concurrent.Executors;

OpenAIClient client = OpenAIOkHttpClient.builder()
    .fromEnv()
    .streamHandlerExecutor(Executors.newFixedThreadPool(4))
    .build();

Binary responses

The SDK defines methods that return binary responses, which are used for API responses that shouldn't necessarily be parsed, like non-JSON data.

These methods return HttpResponse:

import com.openai.core.http.HttpResponse;
import com.openai.models.FileContentParams;

FileContentParams params = FileContentParams.builder()
    .fileId("file_id")
    .build();
HttpResponse response = client.files().content(params);

To save the response content to a file, use the Files.copy(...) method:

import com.openai.core.http.HttpResponse;
import java.nio.file.Files;
import java.nio.file.Paths;
import java.nio.file.StandardCopyOption;

try (HttpResponse response = client.files().content(params)) {
    Files.copy(
        response.body(),
        Paths.get(path),
        StandardCopyOption.REPLACE_EXISTING
    );
} catch (Exception e) {
    System.out.println("Something went wrong!");
    throw new RuntimeException(e);
}

Or transfer the response content to any OutputStream:

import com.openai.core.http.HttpResponse;
import java.nio.file.Files;
import java.nio.file.Paths;

try (HttpResponse response = client.files().content(params)) {
    response.body().transferTo(Files.newOutputStream(Paths.get(path)));
} catch (Exception e) {
    System.out.println("Something went wrong!");
    throw new RuntimeException(e);
}

Error handling

The SDK throws custom unchecked exception types:

• OpenAIServiceException: Base class for HTTP errors. See this table for which exception subclass is thrown for each HTTP status code:

  Status	Exception
  400	BadRequestException
  401	AuthenticationException
  403	PermissionDeniedException
  404	NotFoundException
  422	UnprocessableEntityException
  429	RateLimitException
  5xx	InternalServerException
  others	UnexpectedStatusCodeException

• OpenAIIoException: I/O networking errors.

• OpenAIInvalidDataException: Failure to interpret successfully parsed data. For example, when accessing a property that's supposed to be required, but the API unexpectedly omitted it from the response.

• OpenAIException: Base class for all exceptions. Most errors will result in one of the previously mentioned ones, but completely generic errors may be thrown using the base class.

Pagination

For methods that return a paginated list of results, this library provides convenient ways access the results either one page at a time, or item-by-item across all pages.

Auto-pagination

To iterate through all results across all pages, you can use autoPager, which automatically handles fetching more pages for you:

Synchronous

import com.openai.models.FineTuningJob;
import com.openai.models.FineTuningJobListPage;

// As an Iterable:
FineTuningJobListPage page = client.fineTuning().jobs().list(params);
for (FineTuningJob job : page.autoPager()) {
    System.out.println(job);
};

// As a Stream:
client.fineTuning().jobs().list(params).autoPager().stream()
    .limit(50)
    .forEach(job -> System.out.println(job));

Asynchronous

// Using forEach, which returns CompletableFuture<Void>:
asyncClient.fineTuning().jobs().list(params).autoPager()
    .forEach(job -> System.out.println(job), executor);

Manual pagination

If none of the above helpers meet your needs, you can also manually request pages one-by-one. A page of results has a data() method to fetch the list of objects, as well as top-level response and other methods to fetch top-level data about the page. It also has methods hasNextPage, getNextPage, and getNextPageParams methods to help with pagination.

import com.openai.models.FineTuningJob;
import com.openai.models.FineTuningJobListPage;

FineTuningJobListPage page = client.fineTuning().jobs().list(params);
while (page != null) {
    for (FineTuningJob job : page.data()) {
        System.out.println(job);
    }

    page = page.getNextPage().orElse(null);
}

Logging

The SDK uses the standard OkHttp logging interceptor.

Enable logging by setting the OPENAI_LOG environment variable to info:

$ export OPENAI_LOG=info

Or to debug for more verbose logging:

$ export OPENAI_LOG=debug

Microsoft Azure

To use this library with Azure OpenAI, use the same OpenAI client builder but with the Azure-specific configuration.

OpenAIClient client = OpenAIOkHttpClient.builder()
        // Gets the API key from the `AZURE_OPENAI_KEY` environment variable
        .fromEnv()
        // Set the Azure Entra ID
        .credential(BearerTokenCredential.create(AuthenticationUtil.getBearerTokenSupplier(
                new DefaultAzureCredentialBuilder().build(), "https://cognitiveservices.azure.com/.default")))
        .build();

See the complete Azure OpenAI example in the openai-java-example directory. The other examples in the directory also work with Azure as long as the client is configured to use it.

Network options

Retries

The SDK automatically retries 2 times by default, with a short exponential backoff.

Only the following error types are retried:

• Connection errors (for example, due to a network connectivity problem)
• 408 Request Timeout
• 409 Conflict
• 429 Rate Limit
• 5xx Internal

The API may also explicitly instruct the SDK to retry or not retry a response.

To set a custom number of retries, configure the client using the maxRetries method:

import com.openai.client.OpenAIClient;
import com.openai.client.okhttp.OpenAIOkHttpClient;

OpenAIClient client = OpenAIOkHttpClient.builder()
    .fromEnv()
    .maxRetries(4)
    .build();

Timeouts

Requests time out after 10 minutes by default.

To set a custom timeout, configure the method call using the timeout method:

import com.openai.models.ChatCompletion;
import com.openai.models.ChatCompletionCreateParams;
import com.openai.models.ChatModel;

ChatCompletion chatCompletion = client.chat().completions().create(
  params, RequestOptions.builder().timeout(Duration.ofSeconds(30)).build()
);

Or configure the default for all method calls at the client level:

import com.openai.client.OpenAIClient;
import com.openai.client.okhttp.OpenAIOkHttpClient;
import java.time.Duration;

OpenAIClient client = OpenAIOkHttpClient.builder()
    .fromEnv()
    .timeout(Duration.ofSeconds(30))
    .build();

Proxies

To route requests through a proxy, configure the client using the proxy method:

import com.openai.client.OpenAIClient;
import com.openai.client.okhttp.OpenAIOkHttpClient;
import java.net.InetSocketAddress;
import java.net.Proxy;

OpenAIClient client = OpenAIOkHttpClient.builder()
    .fromEnv()
    .proxy(new Proxy(
      Proxy.Type.HTTP, new InetSocketAddress(
        "https://example.com", 8080
      )
    ))
    .build();

Undocumented API functionality

The SDK is typed for convenient usage of the documented API. However, it also supports working with undocumented or not yet supported parts of the API.

Parameters

To set undocumented parameters, call the putAdditionalHeader, putAdditionalQueryParam, or putAdditionalBodyProperty methods on any Params class:

import com.openai.core.JsonValue;
import com.openai.models.ChatCompletionCreateParams;

ChatCompletionCreateParams params = ChatCompletionCreateParams.builder()
    .putAdditionalHeader("Secret-Header", "42")
    .putAdditionalQueryParam("secret_query_param", "42")
    .putAdditionalBodyProperty("secretProperty", JsonValue.from("42"))
    .build();

These can be accessed on the built object later using the _additionalHeaders(), _additionalQueryParams(), and _additionalBodyProperties() methods. You can also set undocumented parameters on nested headers, query params, or body classes using the putAdditionalProperty method. These properties can be accessed on the built object later using the _additionalProperties() method.

To set a documented parameter or property to an undocumented or not yet supported value, pass a JsonValue object to its setter:

import com.openai.core.JsonValue;
import com.openai.models.ChatCompletionCreateParams;

ChatCompletionCreateParams params = ChatCompletionCreateParams.builder()
    .addUserMessage("Say this is a test")
    .model(JsonValue.from(42))
    .build();

Response properties

To access undocumented response properties, call the _additionalProperties() method:

import com.openai.core.JsonValue;
import java.util.Map;

Map<String, JsonValue> additionalProperties = client.chat().completions().create(params)._additionalProperties();
JsonValue secretPropertyValue = additionalProperties.get("secretProperty");

String result = secretPropertyValue.accept(new JsonValue.Visitor<>() {
    @Override
    public String visitNull() {
        return "It's null!";
    }

    @Override
    public String visitBoolean(boolean value) {
        return "It's a boolean!";
    }

    @Override
    public String visitNumber(Number value) {
        return "It's a number!";
    }

    // Other methods include `visitMissing`, `visitString`, `visitArray`, and `visitObject`
    // The default implementation of each unimplemented method delegates to `visitDefault`, which throws by default, but can also be overridden
});

To access a property's raw JSON value, which may be undocumented, call its _ prefixed method:

import com.openai.core.JsonField;
import com.openai.models.ChatCompletionMessageParam;
import java.util.Optional;

JsonField<List<ChatCompletionMessageParam>> messages = client.chat().completions().create(params)._messages();

if (messages.isMissing()) {
  // The property is absent from the JSON response
} else if (messages.isNull()) {
  // The property was set to literal null
} else {
  // Check if value was provided as a string
  // Other methods include `asNumber()`, `asBoolean()`, etc.
  Optional<String> jsonString = messages.asString();

  // Try to deserialize into a custom type
  MyClass myObject = messages.asUnknown().orElseThrow().convert(MyClass.class);
}

Response validation

In rare cases, the API may return a response that doesn't match the expected type. For example, the SDK may expect a property to contain a String, but the API could return something else.

By default, the SDK will not throw an exception in this case. It will throw OpenAIInvalidDataException only if you directly access the property.

If you would prefer to check that the response is completely well-typed upfront, then either call validate():

import com.openai.models.ChatCompletion;

ChatCompletion chatCompletion = client.chat().completions().create(params).validate();

Or configure the method call to validate the response using the responseValidation method:

import com.openai.models.ChatCompletion;
import com.openai.models.ChatCompletionCreateParams;
import com.openai.models.ChatModel;

ChatCompletion chatCompletion = client.chat().completions().create(
  params, RequestOptions.builder().responseValidation(true).build()
);

Or configure the default for all method calls at the client level:

import com.openai.client.OpenAIClient;
import com.openai.client.okhttp.OpenAIOkHttpClient;

OpenAIClient client = OpenAIOkHttpClient.builder()
    .fromEnv()
    .responseValidation(true)
    .build();

FAQ

Why don't you use plain enum classes?

Java enum classes are not trivially forwards compatible. Using them in the SDK could cause runtime exceptions if the API is updated to respond with a new enum value.

Why do you represent fields using JsonField<T> instead of just plain T?

Using JsonField<T> enables a few features:

• Allowing usage of undocumented API functionality
• Lazily validating the API response against the expected shape
• Representing absent vs explicitly null values

Why don't you use data classes?

It is not backwards compatible to add new fields to a data class and we don't want to introduce a breaking change every time we add a field to a class.

Why don't you use checked exceptions?

Checked exceptions are widely considered a mistake in the Java programming language. In fact, they were omitted from Kotlin for this reason.

Checked exceptions:

• Are verbose to handle
• Encourage error handling at the wrong level of abstraction, where nothing can be done about the error
• Are tedious to propagate due to the function coloring problem
• Don't play well with lambdas (also due to the function coloring problem)

Semantic versioning

This package generally follows SemVer conventions, though certain backwards-incompatible changes may be released as minor versions:

1. Changes to library internals which are technically public but not intended or documented for external use. (Please open a GitHub issue to let us know if you are relying on such internals.)
2. Changes that we do not expect to impact the vast majority of users in practice.

We take backwards-compatibility seriously and work hard to ensure you can rely on a smooth upgrade experience.

We are keen for your feedback; please open an issue with questions, bugs, or suggestions.