Class AnthropicStreamingChatModel.AnthropicStreamingChatModelBuilder

java.lang.Object

dev.langchain4j.model.anthropic.AnthropicStreamingChatModel.AnthropicStreamingChatModelBuilder

Enclosing class:

AnthropicStreamingChatModel

public static class AnthropicStreamingChatModel.AnthropicStreamingChatModelBuilder
extends Object

Constructor Summary

Constructors

Constructor	Description
AnthropicStreamingChatModelBuilder()

Method Summary

Modifier and Type	Method	Description
AnthropicStreamingChatModel.AnthropicStreamingChatModelBuilder	apiKey(String apiKey)	Sets the Anthropic API key used to authenticate requests.
AnthropicStreamingChatModel.AnthropicStreamingChatModelBuilder	baseUrl(String baseUrl)	Sets the base URL of the Anthropic API.
AnthropicStreamingChatModel.AnthropicStreamingChatModelBuilder	beta(String beta)	Sets the value of the anthropic-beta request header to opt into beta features.
AnthropicStreamingChatModel	build()
AnthropicStreamingChatModel.AnthropicStreamingChatModelBuilder	cacheAutomatically(Boolean cacheAutomatically)	Enables automatic prompt caching of the conversation.
AnthropicStreamingChatModel.AnthropicStreamingChatModelBuilder	cacheSystemMessages(Boolean cacheSystemMessages)	Enables prompt caching for SystemMessages.
AnthropicStreamingChatModel.AnthropicStreamingChatModelBuilder	cacheTools(Boolean cacheTools)	Enables prompt caching for ToolSpecifications.
AnthropicStreamingChatModel.AnthropicStreamingChatModelBuilder	cacheTtl(String cacheTtl)	Sets how long the cached prompt is kept: AnthropicChatRequestParameters.CACHE_TTL_5M (5 minutes, the default) or AnthropicChatRequestParameters.CACHE_TTL_1H (1 hour).
AnthropicStreamingChatModel.AnthropicStreamingChatModelBuilder	customHeaders(Supplier<Map<String,String>> customHeadersSupplier)	Sets a supplier of extra HTTP headers to include in every request to the Anthropic API.
AnthropicStreamingChatModel.AnthropicStreamingChatModelBuilder	customHeaders(Map<String,String> customHeaders)	Sets extra HTTP headers to include in every request to the Anthropic API.
AnthropicStreamingChatModel.AnthropicStreamingChatModelBuilder	customParameters(Map<String,Object> customParameters)	Sets arbitrary extra parameters to include in the Anthropic API request body.
AnthropicStreamingChatModel.AnthropicStreamingChatModelBuilder	defaultRequestParameters(ChatRequestParameters parameters)	Sets default ChatRequestParameters that are merged into every request.
AnthropicStreamingChatModel.AnthropicStreamingChatModelBuilder	disableParallelToolUse(Boolean disableParallelToolUse)	When set to true, prevents the model from calling multiple tools in a single response turn.
AnthropicStreamingChatModel.AnthropicStreamingChatModelBuilder	httpClientBuilder(HttpClientBuilder httpClientBuilder)	Sets a custom HttpClientBuilder for the underlying HTTP client.
AnthropicStreamingChatModel.AnthropicStreamingChatModelBuilder	listeners(ChatModelListener... listeners)	Sets the ChatModelListeners to be notified on each request and response.
AnthropicStreamingChatModel.AnthropicStreamingChatModelBuilder	listeners(List<ChatModelListener> listeners)	Sets the list of ChatModelListeners to be notified on each request and response.
AnthropicStreamingChatModel.AnthropicStreamingChatModelBuilder	logger(org.slf4j.Logger logger)
AnthropicStreamingChatModel.AnthropicStreamingChatModelBuilder	logRequests(Boolean logRequests)	Enables debug logging of HTTP request bodies sent to the Anthropic API.
AnthropicStreamingChatModel.AnthropicStreamingChatModelBuilder	logResponses(Boolean logResponses)	Enables debug logging of HTTP response bodies (server-sent events) received from the Anthropic API.
AnthropicStreamingChatModel.AnthropicStreamingChatModelBuilder	maxTokens(Integer maxTokens)	Sets the maximum number of tokens to generate in the response.
AnthropicStreamingChatModel.AnthropicStreamingChatModelBuilder	midConversationSystemMessages(Boolean midConversationSystemMessages)	Controls whether a SystemMessage that appears after the conversation has started (a mid-conversation system message) is sent inline as a system entry in the messages array, instead of being merged into the top-level system prompt.
AnthropicStreamingChatModel.AnthropicStreamingChatModelBuilder	modelName(AnthropicChatModelName modelName)	Sets the model to use for streaming chat completions using a type-safe enum constant.
AnthropicStreamingChatModel.AnthropicStreamingChatModelBuilder	modelName(String modelName)	Sets the model to use for streaming chat completions, specified as a string model ID.
AnthropicStreamingChatModel.AnthropicStreamingChatModelBuilder	responseFormat(ResponseFormat responseFormat)	Sets the response format, enabling structured output such as JSON mode.
AnthropicStreamingChatModel.AnthropicStreamingChatModelBuilder	returnCacheDiagnostics(Boolean returnCacheDiagnostics)	Enables Anthropic's (beta) cache diagnostics for every request.
AnthropicStreamingChatModel.AnthropicStreamingChatModelBuilder	returnServerToolResults(Boolean returnServerToolResults)	Controls whether to return server tool results (e.g., web_search, code_execution) inside AiMessage.attributes() under the key "server_tool_results".
AnthropicStreamingChatModel.AnthropicStreamingChatModelBuilder	returnThinking(Boolean returnThinking)	Controls whether to return thinking/reasoning text (if available) inside AiMessage.thinking() and whether to invoke the StreamingChatResponseHandler.onPartialThinking(PartialThinking) callback.
AnthropicStreamingChatModel.AnthropicStreamingChatModelBuilder	sendThinking(Boolean sendThinking)	Controls whether to send thinking/reasoning text to the LLM in follow-up requests.
AnthropicStreamingChatModel.AnthropicStreamingChatModelBuilder	serverTools(AnthropicServerTool... serverTools)	Specifies server tools to be included in each request to the Anthropic API.
AnthropicStreamingChatModel.AnthropicStreamingChatModelBuilder	serverTools(List<AnthropicServerTool> serverTools)	Specifies server tools to be included in each request to the Anthropic API.
AnthropicStreamingChatModel.AnthropicStreamingChatModelBuilder	skills(AnthropicSkill... skills)
AnthropicStreamingChatModel.AnthropicStreamingChatModelBuilder	skills(List<AnthropicSkill> skills)	Enables Anthropic Agent Skills so Claude can generate real downloadable documents (e.g.
AnthropicStreamingChatModel.AnthropicStreamingChatModelBuilder	stopSequences(List<String> stopSequences)	Sets sequences that, when generated, will cause the model to stop generating further tokens.
AnthropicStreamingChatModel.AnthropicStreamingChatModelBuilder	streamingBufferSize(Integer streamingBufferSize)	Sets the size of the bounded back-pressure buffer for the reactive (Flow.Publisher) streaming path.
AnthropicStreamingChatModel.AnthropicStreamingChatModelBuilder	strictTools(Boolean strictTools)	Enables strict JSON schema validation for tool input parameters.
AnthropicStreamingChatModel.AnthropicStreamingChatModelBuilder	supportedCapabilities(Capability... supportedCapabilities)	Declares the capabilities supported by the model (e.g.
AnthropicStreamingChatModel.AnthropicStreamingChatModelBuilder	supportedCapabilities(Set<Capability> supportedCapabilities)	Declares the capabilities supported by the model.
AnthropicStreamingChatModel.AnthropicStreamingChatModelBuilder	temperature(Double temperature)	Sets the sampling temperature in the range [0.0, 1.0].
AnthropicStreamingChatModel.AnthropicStreamingChatModelBuilder	thinkingBudgetTokens(Integer thinkingBudgetTokens)	Configures thinking.
AnthropicStreamingChatModel.AnthropicStreamingChatModelBuilder	thinkingDisplay(String thinkingDisplay)	Controls whether the API streams readable thinking text next to the thinking signature.
AnthropicStreamingChatModel.AnthropicStreamingChatModelBuilder	thinkingType(String thinkingType)	Enables thinking.
AnthropicStreamingChatModel.AnthropicStreamingChatModelBuilder	timeout(Duration timeout)	Sets the HTTP request timeout for calls to the Anthropic streaming API.
AnthropicStreamingChatModel.AnthropicStreamingChatModelBuilder	toolChoice(ToolChoice toolChoice)	Controls how the model uses tools.
AnthropicStreamingChatModel.AnthropicStreamingChatModelBuilder	toolChoiceName(String toolChoiceName)	Sets the name of the tool that the model is forced to call.
AnthropicStreamingChatModel.AnthropicStreamingChatModelBuilder	toolMetadataKeysToSend(String... toolMetadataKeysToSend)	Specifies metadata keys from the ToolSpecification.metadata() to be included in the request.
AnthropicStreamingChatModel.AnthropicStreamingChatModelBuilder	toolMetadataKeysToSend(Set<String> toolMetadataKeysToSend)	Specifies metadata keys from the ToolSpecification.metadata() to be included in the request.
AnthropicStreamingChatModel.AnthropicStreamingChatModelBuilder	toolSpecifications(ToolSpecification... toolSpecifications)	Sets the tools (functions) available to the model for function calling.
AnthropicStreamingChatModel.AnthropicStreamingChatModelBuilder	toolSpecifications(List<ToolSpecification> toolSpecifications)	Sets the list of tools (functions) available to the model for function calling.
AnthropicStreamingChatModel.AnthropicStreamingChatModelBuilder	topK(Integer topK)	Sets the top-K sampling value.
AnthropicStreamingChatModel.AnthropicStreamingChatModelBuilder	topP(Double topP)	Sets the nucleus sampling probability (top-p) in the range (0.0, 1.0].
AnthropicStreamingChatModel.AnthropicStreamingChatModelBuilder	userId(String userId)	Sets the user ID for the requests.
AnthropicStreamingChatModel.AnthropicStreamingChatModelBuilder	version(String version)	Sets the value of the anthropic-version request header.

Methods inherited from class Object

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

Constructor Details

AnthropicStreamingChatModelBuilder

public AnthropicStreamingChatModelBuilder()

Method Details

httpClientBuilder

public AnthropicStreamingChatModel.AnthropicStreamingChatModelBuilder httpClientBuilder(HttpClientBuilder httpClientBuilder)

Sets a custom HttpClientBuilder for the underlying HTTP client. Use this to configure timeouts, proxies, or other HTTP-level settings.

Parameters:

httpClientBuilder - the HTTP client builder

Returns:

this

streamingBufferSize

@Experimental
public AnthropicStreamingChatModel.AnthropicStreamingChatModelBuilder streamingBufferSize(Integer streamingBufferSize)

Sets the size of the bounded back-pressure buffer for the reactive (Flow.Publisher) streaming path. Events from the model are relayed through this buffer; if a subscriber consumes slower than the model produces and the buffer overflows, the stream terminates with an IllegalStateException. Defaults to 16384.

Parameters:

streamingBufferSize - the buffer size (must be greater than zero)

Returns:

this

Since:

1.20.0

baseUrl

public AnthropicStreamingChatModel.AnthropicStreamingChatModelBuilder baseUrl(String baseUrl)

Sets the base URL of the Anthropic API.

Defaults to https://api.anthropic.com/v1/.

Parameters:

baseUrl - the base URL

Returns:

this

apiKey

public AnthropicStreamingChatModel.AnthropicStreamingChatModelBuilder apiKey(String apiKey)

Sets the Anthropic API key used to authenticate requests.

Parameters:

apiKey - the API key

Returns:

this

version

public AnthropicStreamingChatModel.AnthropicStreamingChatModelBuilder version(String version)

Sets the value of the anthropic-version request header.

Defaults to 2023-06-01. See the Anthropic API versioning docs.

Parameters:

version - the API version string

Returns:

this

beta

public AnthropicStreamingChatModel.AnthropicStreamingChatModelBuilder beta(String beta)

Sets the value of the anthropic-beta request header to opt into beta features.

See the Anthropic beta headers docs.

Parameters:

beta - the beta feature identifier

Returns:

this

modelName

public AnthropicStreamingChatModel.AnthropicStreamingChatModelBuilder modelName(String modelName)

Sets the model to use for streaming chat completions, specified as a string model ID.

See AnthropicChatModelName for available model constants.

Parameters:

modelName - the model ID, e.g. "claude-opus-4-5"

Returns:

this

modelName

public AnthropicStreamingChatModel.AnthropicStreamingChatModelBuilder modelName(AnthropicChatModelName modelName)

Sets the model to use for streaming chat completions using a type-safe enum constant.

Parameters:

modelName - the model name enum value

Returns:

this

temperature

public AnthropicStreamingChatModel.AnthropicStreamingChatModelBuilder temperature(Double temperature)

Sets the sampling temperature in the range [0.0, 1.0]. Higher values produce more random output; lower values produce more deterministic output.

Cannot be used together with topP(Double).

Parameters:

temperature - the sampling temperature

Returns:

this

topP

public AnthropicStreamingChatModel.AnthropicStreamingChatModelBuilder topP(Double topP)

Sets the nucleus sampling probability (top-p) in the range (0.0, 1.0]. Only the tokens whose cumulative probability exceeds this threshold are considered.

Cannot be used together with temperature(Double).

Parameters:

topP - the nucleus sampling threshold

Returns:

this

topK

public AnthropicStreamingChatModel.AnthropicStreamingChatModelBuilder topK(Integer topK)

Sets the top-K sampling value. Only the topK most-likely next tokens are considered at each step.

Recommended for advanced use only; temperature(Double) is usually sufficient.

Parameters:

topK - the number of top tokens to sample from

Returns:

this

maxTokens

public AnthropicStreamingChatModel.AnthropicStreamingChatModelBuilder maxTokens(Integer maxTokens)

Sets the maximum number of tokens to generate in the response.

Defaults to 1024 if not set.

Parameters:

maxTokens - the maximum number of output tokens

Returns:

this

stopSequences

public AnthropicStreamingChatModel.AnthropicStreamingChatModelBuilder stopSequences(List<String> stopSequences)

Sets sequences that, when generated, will cause the model to stop generating further tokens.

Parameters:

stopSequences - the list of stop sequences

Returns:

this

responseFormat

public AnthropicStreamingChatModel.AnthropicStreamingChatModelBuilder responseFormat(ResponseFormat responseFormat)

Sets the response format, enabling structured output such as JSON mode.

Parameters:

responseFormat - the desired response format

Returns:

this

toolSpecifications

public AnthropicStreamingChatModel.AnthropicStreamingChatModelBuilder toolSpecifications(List<ToolSpecification> toolSpecifications)

Sets the list of tools (functions) available to the model for function calling.

Parameters:

toolSpecifications - the tool specifications

Returns:

this

toolSpecifications

public AnthropicStreamingChatModel.AnthropicStreamingChatModelBuilder toolSpecifications(ToolSpecification... toolSpecifications)

Sets the tools (functions) available to the model for function calling.

Parameters:

toolSpecifications - the tool specifications

Returns:

this

cacheSystemMessages

public AnthropicStreamingChatModel.AnthropicStreamingChatModelBuilder cacheSystemMessages(Boolean cacheSystemMessages)

Enables prompt caching for SystemMessages.

When true, the last system message is marked with cache_control, so that Anthropic caches the tools and system messages across requests, reducing cost and latency for repeated prompts. It works for any kind of usage, including independent calls without chat memory. To also cache a conversation whose history grows from one request to the next, enable cacheAutomatically as well. See the prompt caching docs.

Parameters:

cacheSystemMessages - whether to cache system messages

Returns:

this

cacheTools

public AnthropicStreamingChatModel.AnthropicStreamingChatModelBuilder cacheTools(Boolean cacheTools)

Enables prompt caching for ToolSpecifications.

When true, the last tool definition is marked with cache_control, so that Anthropic caches the tool definitions across requests. To also cache a conversation whose history grows from one request to the next, enable cacheAutomatically as well. See the prompt caching docs.

Parameters:

cacheTools - whether to cache tool definitions

Returns:

this

cacheAutomatically

public AnthropicStreamingChatModel.AnthropicStreamingChatModelBuilder cacheAutomatically(Boolean cacheAutomatically)

Enables automatic prompt caching of the conversation.

When true, Anthropic places a cache breakpoint on the last cacheable block of each request and moves it forward as the conversation grows, so that every request reads the whole previous conversation (tools, system messages and earlier messages) from the cache instead of processing it again. No message has to be marked for caching by hand, so it also works for AI Services and agents.

It only pays off when each request starts with everything the previous request sent, for example when an AI Service or an agent calls tools in a loop, or in a chat whose memory still holds all previous messages. When the beginning of the conversation changes on every request (for example, when a chat memory such as a full MessageWindowChatMemory evicts old messages on every turn, or for independent calls without chat memory), every request pays the cache write price and nothing is read back, which costs more than not caching at all.

When enabling it, enable cacheSystemMessages and cacheTools as well, to keep the system messages and tools cached even when the beginning of the conversation changes. Each of these options, as well as each message marked with the cache_control attribute, uses one of the 4 cache breakpoints Anthropic allows per request; a request with more is rejected.

This option is sent as a top-level cache_control field of the request. Anthropic-compatible gateways and proxies that do not support this field may reject the request or ignore the field.

Writing to the cache costs more than regular input tokens, and reading from it costs much less, so caching pays off when the same prefix is sent more than once within the cache TTL (see cacheTtl). Prompts shorter than the model-specific minimum are not cached. Disabled by default. See the prompt caching docs.

Parameters:

cacheAutomatically - whether to enable automatic caching of the conversation

Returns:

this

cacheTtl

public AnthropicStreamingChatModel.AnthropicStreamingChatModelBuilder cacheTtl(String cacheTtl)

Sets how long the cached prompt is kept: AnthropicChatRequestParameters.CACHE_TTL_5M (5 minutes, the default) or AnthropicChatRequestParameters.CACHE_TTL_1H (1 hour).

Applies to every cache breakpoint: system messages, tools, automatic conversation caching and messages marked with the cache_control attribute. It has no effect unless at least one of them is cached. The value is sent to Anthropic as is; a blank value is rejected. Each cache hit refreshes the TTL, so "5m" is enough when requests sharing the same prefix are less than 5 minutes apart. Use "1h" when they are further apart (for example, a user replying after 20 minutes, or batch processing). Writing to the 1-hour cache costs more than writing to the 5-minute one. See the prompt caching docs.

Parameters:

cacheTtl - the cache TTL, AnthropicChatRequestParameters.CACHE_TTL_5M or AnthropicChatRequestParameters.CACHE_TTL_1H

Returns:

this

thinkingType

public AnthropicStreamingChatModel.AnthropicStreamingChatModelBuilder thinkingType(String thinkingType)

Enables thinking.

thinkingBudgetTokens

public AnthropicStreamingChatModel.AnthropicStreamingChatModelBuilder thinkingBudgetTokens(Integer thinkingBudgetTokens)

Configures thinking.

thinkingDisplay

public AnthropicStreamingChatModel.AnthropicStreamingChatModelBuilder thinkingDisplay(String thinkingDisplay)

Controls whether the API streams readable thinking text next to the thinking signature.

Valid values:

• "summarized": thinking blocks contain a readable summary of the reasoning.
• "omitted": thinking blocks contain an empty thinking text, only the encrypted signature is returned.

When this is not set, the API picks a default that depends on the model: recent Claude models default to "omitted", older ones to "summarized". Set it to "summarized" whenever the thinking text itself is needed, for example in order to stream it to the end user.

The model thinks and is billed the same way in both cases; only the visibility of the thinking text changes.

See Also:

• Anthropic documentation
• thinkingType(String)
• returnThinking(Boolean)

returnThinking

public AnthropicStreamingChatModel.AnthropicStreamingChatModelBuilder returnThinking(Boolean returnThinking)

Controls whether to return thinking/reasoning text (if available) inside AiMessage.thinking() and whether to invoke the StreamingChatResponseHandler.onPartialThinking(PartialThinking) callback. Please note that this does not enable thinking/reasoning for the LLM; it only controls whether to parse the thinking field from the API response and return it inside the AiMessage.

Disabled by default. If enabled, the thinking text will be stored within the AiMessage and may be persisted. If enabled, thinking signatures will also be stored and returned inside the AiMessage.attributes().

Please note that AiMessage.thinking() stays empty and StreamingChatResponseHandler.onPartialThinking(PartialThinking) is not invoked when the API returns no thinking text, which is the default for recent Claude models. See thinkingDisplay(String).

See Also:

• thinkingType(String)
• thinkingBudgetTokens(Integer)
• thinkingDisplay(String)
• sendThinking(Boolean)

sendThinking

public AnthropicStreamingChatModel.AnthropicStreamingChatModelBuilder sendThinking(Boolean sendThinking)

Controls whether to send thinking/reasoning text to the LLM in follow-up requests.

Enabled by default. If enabled, the contents of AiMessage.thinking() will be sent in the API request. If enabled, thinking signatures (inside the AiMessage.attributes()) will also be sent.

See Also:

• thinkingType(String)
• thinkingBudgetTokens(Integer)
• returnThinking(Boolean)

midConversationSystemMessages

public AnthropicStreamingChatModel.AnthropicStreamingChatModelBuilder midConversationSystemMessages(Boolean midConversationSystemMessages)

Controls whether a SystemMessage that appears after the conversation has started (a mid-conversation system message) is sent inline as a system entry in the messages array, instead of being merged into the top-level system prompt.

Disabled by default, in which case every SystemMessage is sent via the top-level system prompt regardless of position (the existing behavior). When enabled, leading SystemMessages still populate the top-level system prompt, while later ones are sent inline so they take effect from that point in the conversation onward.

Supported only by Claude Opus 4.8. Anthropic also constrains placement: a mid-conversation system message must immediately follow a user turn (including a user turn carrying tool results) and must not sit between a tool_use block and its tool_result; an unsupported model or an invalid placement returns a 400. This library does not reorder messages; it sends them at the position the caller provided.

Parameters:

midConversationSystemMessages - whether to send mid-conversation system messages inline

Returns:

this

See Also:

• Anthropic: mid-conversation system messages

timeout

public AnthropicStreamingChatModel.AnthropicStreamingChatModelBuilder timeout(Duration timeout)

Sets the HTTP request timeout for calls to the Anthropic streaming API.

Parameters:

timeout - the request timeout

Returns:

this

logRequests

public AnthropicStreamingChatModel.AnthropicStreamingChatModelBuilder logRequests(Boolean logRequests)

Enables debug logging of HTTP request bodies sent to the Anthropic API.

Parameters:

logRequests - whether to log requests

Returns:

this

logResponses

public AnthropicStreamingChatModel.AnthropicStreamingChatModelBuilder logResponses(Boolean logResponses)

Enables debug logging of HTTP response bodies (server-sent events) received from the Anthropic API.

Parameters:

logResponses - whether to log responses

Returns:

this

logger

public AnthropicStreamingChatModel.AnthropicStreamingChatModelBuilder logger(org.slf4j.Logger logger)

Parameters:

logger - an alternate Logger to be used instead of the default one provided by Langchain4J for logging requests and responses.

Returns:

this.

listeners

public AnthropicStreamingChatModel.AnthropicStreamingChatModelBuilder listeners(List<ChatModelListener> listeners)

Sets the list of ChatModelListeners to be notified on each request and response. Useful for logging, metrics, and observability integrations.

Parameters:

listeners - the chat model listeners

Returns:

this

listeners

public AnthropicStreamingChatModel.AnthropicStreamingChatModelBuilder listeners(ChatModelListener... listeners)

Sets the ChatModelListeners to be notified on each request and response.

Parameters:

listeners - the chat model listeners

Returns:

this

defaultRequestParameters

public AnthropicStreamingChatModel.AnthropicStreamingChatModelBuilder defaultRequestParameters(ChatRequestParameters parameters)

Sets default ChatRequestParameters that are merged into every request. Individual request parameters take precedence over these defaults.

Parameters:

parameters - the default request parameters

Returns:

this

toolChoice

public AnthropicStreamingChatModel.AnthropicStreamingChatModelBuilder toolChoice(ToolChoice toolChoice)

Controls how the model uses tools.

Use ToolChoice.AUTO to let the model decide, ToolChoice.REQUIRED to force tool use, or ToolChoice.NONE to disable tool use.

Parameters:

toolChoice - the tool choice strategy

Returns:

this

toolChoiceName

public AnthropicStreamingChatModel.AnthropicStreamingChatModelBuilder toolChoiceName(String toolChoiceName)

Sets the name of the tool that the model is forced to call.

It can be set on its own; it does not require toolChoice(ToolChoice). When toolChoice(ToolChoice) is set as well, the named tool takes precedence over it. It has no effect when the request contains no tools.

Parameters:

toolChoiceName - the name of the tool to force

Returns:

this

disableParallelToolUse

public AnthropicStreamingChatModel.AnthropicStreamingChatModelBuilder disableParallelToolUse(Boolean disableParallelToolUse)

When set to true, prevents the model from calling multiple tools in a single response turn.

Parameters:

disableParallelToolUse - whether to disable parallel tool use

Returns:

this

serverTools

public AnthropicStreamingChatModel.AnthropicStreamingChatModelBuilder serverTools(List<AnthropicServerTool> serverTools)

Specifies server tools to be included in each request to the Anthropic API. For example:

AnthropicServerTool webSearchTool = AnthropicServerTool.builder()
    .type("web_search_20250305")
    .name("web_search")
    .addAttribute("max_uses", 5)
    .addAttribute("allowed_domains", List.of("accuweather.com"))
    .build();

serverTools

public AnthropicStreamingChatModel.AnthropicStreamingChatModelBuilder serverTools(AnthropicServerTool... serverTools)

Specifies server tools to be included in each request to the Anthropic API. For example:

AnthropicServerTool webSearchTool = AnthropicServerTool.builder()
    .type("web_search_20250305")
    .name("web_search")
    .addAttribute("max_uses", 5)
    .addAttribute("allowed_domains", List.of("accuweather.com"))
    .build();

skills

public AnthropicStreamingChatModel.AnthropicStreamingChatModelBuilder skills(List<AnthropicSkill> skills)

Enables Anthropic Agent Skills so Claude can generate real downloadable documents (e.g. .xlsx, .pptx, .docx, .pdf).

Enabling skills automatically adds the container.skills block and the code_execution server tool (unless already configured via serverTools(List)), so that does not need to be wired up manually.

You must, however, opt into the required beta features yourself via beta(String), for example .beta("code-execution-2025-08-25,skills-2025-10-02,files-api-2025-04-14"). These are beta headers and their values change over time; see the Agent Skills docs for the current set.

Combine with returnServerToolResults(Boolean) to surface the generated file ids under the "server_tool_results" key of AiMessage.attributes().

Skills are supported on Claude Sonnet 4 / 4.5, Opus 4 and later. At most 8 skills may be enabled per request.

skills

public AnthropicStreamingChatModel.AnthropicStreamingChatModelBuilder skills(AnthropicSkill... skills)

See Also:

• skills(List)

returnServerToolResults

public AnthropicStreamingChatModel.AnthropicStreamingChatModelBuilder returnServerToolResults(Boolean returnServerToolResults)

Controls whether to return server tool results (e.g., web_search, code_execution) inside AiMessage.attributes() under the key "server_tool_results".

Disabled by default to avoid polluting ChatMemory with potentially large data. If enabled, server tool results will be stored as a List<AnthropicServerToolResult> within the AiMessage attributes.

See Also:

• serverTools(List)

toolMetadataKeysToSend

public AnthropicStreamingChatModel.AnthropicStreamingChatModelBuilder toolMetadataKeysToSend(Set<String> toolMetadataKeysToSend)

Specifies metadata keys from the ToolSpecification.metadata() to be included in the request.

toolMetadataKeysToSend

public AnthropicStreamingChatModel.AnthropicStreamingChatModelBuilder toolMetadataKeysToSend(String... toolMetadataKeysToSend)

Specifies metadata keys from the ToolSpecification.metadata() to be included in the request.

userId

public AnthropicStreamingChatModel.AnthropicStreamingChatModelBuilder userId(String userId)

Sets the user ID for the requests. This should be a uuid, hash value, or other opaque identifier. Anthropic may use this id to help detect abuse. Do not include any identifying information such as name, email address, or phone number.

Parameters:

userId - the user identifier

Returns:

this builder

returnCacheDiagnostics

public AnthropicStreamingChatModel.AnthropicStreamingChatModelBuilder returnCacheDiagnostics(Boolean returnCacheDiagnostics)

Enables Anthropic's (beta) cache diagnostics for every request.

Requires the cache-diagnosis-2026-04-07 beta header to also be set via beta(String). There is intentionally no model-level default for the id to compare against — it must be set per-request via AnthropicChatRequestParameters.Builder.previousMessageId(String); see AnthropicChatRequestParameters.previousMessageId() for why.

See Also:

• AnthropicChatRequestParameters.Builder.returnCacheDiagnostics(Boolean)
• AnthropicCacheDiagnostics

customParameters

public AnthropicStreamingChatModel.AnthropicStreamingChatModelBuilder customParameters(Map<String,Object> customParameters)

Sets arbitrary extra parameters to include in the Anthropic API request body. Use this for experimental or provider-specific fields not yet covered by dedicated builder methods.

Parameters:

customParameters - a map of parameter names to values

Returns:

this

strictTools

public AnthropicStreamingChatModel.AnthropicStreamingChatModelBuilder strictTools(Boolean strictTools)

Enables strict JSON schema validation for tool input parameters. When true, the model enforces the exact schema defined in ToolSpecification.

Parameters:

strictTools - whether to enable strict tool schema validation

Returns:

this

supportedCapabilities

public AnthropicStreamingChatModel.AnthropicStreamingChatModelBuilder supportedCapabilities(Capability... supportedCapabilities)

Declares the capabilities supported by the model (e.g. Capability.RESPONSE_FORMAT_JSON_SCHEMA). This influences how LangChain4j generates requests for this model.

Parameters:

supportedCapabilities - the capabilities to declare

Returns:

this

supportedCapabilities

public AnthropicStreamingChatModel.AnthropicStreamingChatModelBuilder supportedCapabilities(Set<Capability> supportedCapabilities)

Declares the capabilities supported by the model.

Parameters:

supportedCapabilities - the set of capabilities to declare

Returns:

this

customHeaders

public AnthropicStreamingChatModel.AnthropicStreamingChatModelBuilder customHeaders(Map<String,String> customHeaders)

Sets extra HTTP headers to include in every request to the Anthropic API.

Parameters:

customHeaders - a map of header names to values

Returns:

this

customHeaders

public AnthropicStreamingChatModel.AnthropicStreamingChatModelBuilder customHeaders(Supplier<Map<String,String>> customHeadersSupplier)

Sets a supplier of extra HTTP headers to include in every request to the Anthropic API. The supplier is called once per request, allowing dynamic header values.

Parameters:

customHeadersSupplier - a supplier that returns a map of header names to values

Returns:

this

build

public AnthropicStreamingChatModel build()