@langchain/openai 0.5.17 → 0.5.18

package/dist/chat_models.d.ts

@@ -1,65 +1,79 @@
 import { type ClientOptions, OpenAI as OpenAIClient } from "openai";
 import { CallbackManagerForLLMRun } from "@langchain/core/callbacks/manager";
 import { AIMessageChunk, type BaseMessage, ChatMessageChunk, FunctionMessageChunk, HumanMessageChunk, SystemMessageChunk, ToolMessageChunk } from "@langchain/core/messages";
-import { ChatGenerationChunk, type ChatGeneration, type ChatResult } from "@langchain/core/outputs";
+import { ChatGenerationChunk, type ChatResult } from "@langchain/core/outputs";
 import { BaseChatModel, type BindToolsInput, type LangSmithParams, type BaseChatModelParams } from "@langchain/core/language_models/chat_models";
 import { type BaseFunctionCallOptions, type BaseLanguageModelInput, type StructuredOutputMethodOptions } from "@langchain/core/language_models/base";
 import { Runnable } from "@langchain/core/runnables";
-import type { ResponseFormatText, ResponseFormatJSONObject, ResponseFormatJSONSchema } from "openai/resources/shared";
 import { InteropZodType } from "@langchain/core/utils/types";
-import { type OpenAICallOptions, type OpenAIChatInput, type OpenAICoreRequestOptions, type ChatOpenAIResponseFormat, ChatOpenAIReasoningSummary } from "./types.js";
+import { type OpenAICallOptions, type OpenAIChatInput, type OpenAICoreRequestOptions, type ChatOpenAIResponseFormat } from "./types.js";
 import { OpenAIToolChoice } from "./utils/openai.js";
-type ResponsesTool = NonNullable<OpenAIClient.Responses.ResponseCreateParams["tools"]>[number];
-type ResponsesToolChoice = NonNullable<OpenAIClient.Responses.ResponseCreateParams["tool_choice"]>;
-type ChatOpenAIToolType = BindToolsInput | OpenAIClient.Chat.ChatCompletionTool | ResponsesTool;
+export type { OpenAICallOptions, OpenAIChatInput };
+interface TokenUsage {
+    completionTokens?: number;
+    promptTokens?: number;
+    totalTokens?: number;
+}
 interface OpenAILLMOutput {
-    tokenUsage: {
-        completionTokens?: number;
-        promptTokens?: number;
-        totalTokens?: number;
-    };
+    tokenUsage: TokenUsage;
 }
-export type { OpenAICallOptions, OpenAIChatInput };
-export declare function messageToOpenAIRole(message: BaseMessage): OpenAIClient.ChatCompletionRole;
-export declare function _convertMessagesToOpenAIParams(messages: BaseMessage[], model?: string): OpenAIClient.Chat.Completions.ChatCompletionMessageParam[];
-interface BaseChatOpenAICallOptions extends OpenAICallOptions, BaseFunctionCallOptions {
+type OpenAIRoleEnum = "system" | "developer" | "assistant" | "user" | "function" | "tool";
+type OpenAICompletionParam = OpenAIClient.Chat.Completions.ChatCompletionMessageParam;
+export declare function messageToOpenAIRole(message: BaseMessage): OpenAIRoleEnum;
+export declare function _convertMessagesToOpenAIParams(messages: BaseMessage[], model?: string): OpenAICompletionParam[];
+type ExcludeController<T> = T extends {
+    controller: unknown;
+} ? never : T;
+type ExcludeNonController<T> = T extends {
+    controller: unknown;
+} ? T : never;
+type ResponsesCreate = OpenAIClient.Responses["create"];
+type ResponsesCreateParams = Parameters<OpenAIClient.Responses["create"]>[0];
+type ResponsesTool = Exclude<ResponsesCreateParams["tools"], undefined>[number];
+type ResponsesToolChoice = Exclude<ResponsesCreateParams["tool_choice"], undefined>;
+type ResponsesCreateInvoke = ExcludeController<Awaited<ReturnType<ResponsesCreate>>>;
+type ResponsesCreateStream = ExcludeNonController<Awaited<ReturnType<ResponsesCreate>>>;
+type ResponseInvocationParams = Omit<ResponsesCreateParams, "input">;
+type ChatCompletionInvocationParams = Omit<OpenAIClient.Chat.ChatCompletionCreateParams, "messages">;
+type ChatOpenAIToolType = BindToolsInput | OpenAIClient.ChatCompletionTool | ResponsesTool;
+export interface ChatOpenAIStructuredOutputMethodOptions<IncludeRaw extends boolean> extends StructuredOutputMethodOptions<IncludeRaw> {
     /**
-     * A list of tools that the model may use to generate responses.
-     * Each tool can be a function, a built-in tool, or a custom tool definition.
-     * If not provided, the model will not use any tools.
+     * strict: If `true` and `method` = "function_calling", model output is
+     * guaranteed to exactly match the schema. If `true`, the input schema
+     * will also be validated according to
+     * https://platform.openai.com/docs/guides/structured-outputs/supported-schemas.
+     * If `false`, input schema will not be validated and model output will not
+     * be validated.
+     * If `undefined`, `strict` argument will not be passed to the model.
+     *
+     * @version 0.2.6
+     * @note Planned breaking change in version `0.3.0`:
+     * `strict` will default to `true` when `method` is
+     * "function_calling" as of version `0.3.0`.
      */
+    strict?: boolean;
+}
+export interface ChatOpenAICallOptions extends OpenAICallOptions, BaseFunctionCallOptions {
     tools?: ChatOpenAIToolType[];
-    /**
-     * Specifies which tool the model should use to respond.
-     * Can be an {@link OpenAIToolChoice} or a {@link ResponsesToolChoice}.
-     * If not set, the model will decide which tool to use automatically.
-     */
     tool_choice?: OpenAIToolChoice | ResponsesToolChoice;
-    /**
-     * Adds a prompt index to prompts passed to the model to track
-     * what prompt is being used for a given generation.
-     */
     promptIndex?: number;
-    /**
-     * An object specifying the format that the model must output.
-     */
     response_format?: ChatOpenAIResponseFormat;
-    /**
-     * When provided, the completions API will make a best effort to sample
-     * deterministically, such that repeated requests with the same `seed`
-     * and parameters should return the same result.
-     */
     seed?: number;
     /**
      * Additional options to pass to streamed completions.
-     * If provided, this takes precedence over "streamUsage" set at
-     * initialization time.
+     * If provided takes precedence over "streamUsage" set at initialization time.
      */
-    stream_options?: OpenAIClient.Chat.ChatCompletionStreamOptions;
+    stream_options?: {
+        /**
+         * Whether or not to include token usage in the stream.
+         * If set to `true`, this will include an additional
+         * chunk at the end of the stream with the token usage.
+         */
+        include_usage: boolean;
+    };
     /**
-     * The model may choose to call multiple functions in a single turn. You can
-     * set parallel_tool_calls to false which ensures only one tool is called at most.
-     * [Learn more](https://platform.openai.com/docs/guides/function-calling#parallel-function-calling)
+     * Whether or not to restrict the ability to
+     * call multiple tools in one response.
      */
     parallel_tool_calls?: boolean;
     /**
@@ -72,6 +86,8 @@ interface BaseChatOpenAICallOptions extends OpenAICallOptions, BaseFunctionCallO
      * be validated.
      *
      * If `undefined`, `strict` argument will not be passed to the model.
+     *
+     * @version 0.2.6
      */
     strict?: boolean;
     /**
@@ -98,6 +114,13 @@ interface BaseChatOpenAICallOptions extends OpenAICallOptions, BaseFunctionCallO
      * [Learn more](https://platform.openai.com/docs/guides/latency-optimization#use-predicted-outputs).
      */
     prediction?: OpenAIClient.ChatCompletionPredictionContent;
+    /**
+     * Constrains effort on reasoning for reasoning models. Currently supported values are low, medium, and high.
+     * Reducing reasoning effort can result in faster responses and fewer tokens used on reasoning in a response.
+     *
+     * @deprecated Use {@link reasoning} object instead.
+     */
+    reasoning_effort?: OpenAIClient.Chat.ChatCompletionReasoningEffort;
     /**
      * Options for reasoning models.
      *
@@ -107,207 +130,42 @@ interface BaseChatOpenAICallOptions extends OpenAICallOptions, BaseFunctionCallO
      * These options will be ignored when not using a reasoning model.
      */
     reasoning?: OpenAIClient.Reasoning;
-    /**
-     * Service tier to use for this request. Can be "auto", "default", or "flex"
-     * Specifies the service tier for prioritization and latency optimization.
-     */
-    service_tier?: OpenAIClient.Chat.ChatCompletionCreateParams["service_tier"];
-}
-export interface BaseChatOpenAIFields extends Partial<OpenAIChatInput>, BaseChatModelParams {
-    /**
-     * Optional configuration options for the OpenAI client.
-     */
-    configuration?: ClientOptions;
-}
-/** @internal */
-declare abstract class BaseChatOpenAI<CallOptions extends BaseChatOpenAICallOptions> extends BaseChatModel<CallOptions, AIMessageChunk> implements Partial<OpenAIChatInput> {
-    temperature?: number;
-    topP?: number;
-    frequencyPenalty?: number;
-    presencePenalty?: number;
-    n?: number;
-    logitBias?: Record<string, number>;
-    model: string;
-    modelKwargs?: OpenAIChatInput["modelKwargs"];
-    stop?: string[];
-    stopSequences?: string[];
-    user?: string;
-    timeout?: number;
-    streaming: boolean;
-    streamUsage: boolean;
-    maxTokens?: number;
-    logprobs?: boolean;
-    topLogprobs?: number;
-    apiKey?: string;
-    organization?: string;
-    __includeRawResponse?: boolean;
-    protected client: OpenAIClient;
-    protected clientConfig: ClientOptions;
-    /**
-     * Whether the model supports the `strict` argument when passing in tools.
-     * If `undefined` the `strict` argument will not be passed to OpenAI.
-     */
-    supportsStrictToolCalling?: boolean;
-    audio?: OpenAIClient.Chat.ChatCompletionAudioParam;
-    modalities?: Array<OpenAIClient.Chat.ChatCompletionModality>;
-    reasoning?: OpenAIClient.Reasoning;
-    /**
-     * Must be set to `true` in tenancies with Zero Data Retention. Setting to `true` will disable
-     * output storage in the Responses API, but this DOES NOT enable Zero Data Retention in your
-     * OpenAI organization or project. This must be configured directly with OpenAI.
-     *
-     * See:
-     * https://help.openai.com/en/articles/10503543-data-residency-for-the-openai-api
-     * https://platform.openai.com/docs/api-reference/responses/create#responses-create-store
-     *
-     * @default false
-     */
-    zdrEnabled?: boolean | undefined;
-    /**
-     * Service tier to use for this request. Can be "auto", "default", or "flex" or "priority".
-     * Specifies the service tier for prioritization and latency optimization.
-     */
-    service_tier?: OpenAIClient.Chat.ChatCompletionCreateParams["service_tier"];
-    _llmType(): string;
-    static lc_name(): string;
-    get callKeys(): string[];
-    lc_serializable: boolean;
-    get lc_secrets(): {
-        [key: string]: string;
-    } | undefined;
-    get lc_aliases(): Record<string, string>;
-    get lc_serializable_keys(): string[];
-    getLsParams(options: this["ParsedCallOptions"]): LangSmithParams;
-    /** @ignore */
-    _identifyingParams(): Omit<OpenAIClient.Chat.ChatCompletionCreateParams, "messages"> & {
-        model_name: string;
-    } & ClientOptions;
-    /**
-     * Get the identifying parameters for the model
-     */
-    identifyingParams(): Omit<OpenAIClient.Chat.Completions.ChatCompletionCreateParams, "messages"> & {
-        model_name: string;
-    } & ClientOptions;
-    constructor(fields?: BaseChatOpenAIFields);
-    /**
-     * Returns backwards compatible reasoning parameters from constructor params and call options
-     * @internal
-     */
-    protected _getReasoningParams(options?: this["ParsedCallOptions"]): OpenAIClient.Reasoning | undefined;
-    /**
-     * Returns an openai compatible response format from a set of options
-     * @internal
-     */
-    protected _getResponseFormat(resFormat?: CallOptions["response_format"]): ResponseFormatText | ResponseFormatJSONObject | ResponseFormatJSONSchema | undefined;
-    protected _getClientOptions(options: OpenAICoreRequestOptions | undefined): OpenAICoreRequestOptions;
-    protected _convertChatOpenAIToolToCompletionsTool(tool: ChatOpenAIToolType, fields?: {
-        strict?: boolean;
-    }): OpenAIClient.ChatCompletionTool;
-    bindTools(tools: ChatOpenAIToolType[], kwargs?: Partial<CallOptions>): Runnable<BaseLanguageModelInput, AIMessageChunk, CallOptions>;
-    /** @ignore */
-    _combineLLMOutput(...llmOutputs: OpenAILLMOutput[]): OpenAILLMOutput;
-    getNumTokensFromMessages(messages: BaseMessage[]): Promise<{
-        totalCount: number;
-        countPerMessage: number[];
-    }>;
-    /** @internal */
-    protected _getNumTokensFromGenerations(generations: ChatGeneration[]): Promise<number>;
-    /** @internal */
-    protected _getEstimatedTokenCountFromPrompt(messages: BaseMessage[], functions?: OpenAIClient.Chat.ChatCompletionCreateParams.Function[], function_call?: "none" | "auto" | OpenAIClient.Chat.ChatCompletionFunctionCallOption): Promise<number>;
-    withStructuredOutput<RunOutput extends Record<string, any> = Record<string, any>>(outputSchema: InteropZodType<RunOutput> | Record<string, any>, config?: StructuredOutputMethodOptions<false>): Runnable<BaseLanguageModelInput, RunOutput>;
-    withStructuredOutput<RunOutput extends Record<string, any> = Record<string, any>>(outputSchema: InteropZodType<RunOutput> | Record<string, any>, config?: StructuredOutputMethodOptions<true>): Runnable<BaseLanguageModelInput, {
-        raw: BaseMessage;
-        parsed: RunOutput;
-    }>;
-    withStructuredOutput<RunOutput extends Record<string, any> = Record<string, any>>(outputSchema: InteropZodType<RunOutput> | Record<string, any>, config?: StructuredOutputMethodOptions<boolean>): Runnable<BaseLanguageModelInput, RunOutput> | Runnable<BaseLanguageModelInput, {
-        raw: BaseMessage;
-        parsed: RunOutput;
-    }>;
-}
-type ExcludeController<T> = T extends {
-    controller: unknown;
-} ? never : T;
-type ResponsesCreate = OpenAIClient.Responses["create"];
-type ResponsesParse = OpenAIClient.Responses["parse"];
-type ResponsesCreateInvoke = ExcludeController<Awaited<ReturnType<ResponsesCreate>>>;
-type ResponsesParseInvoke = ExcludeController<Awaited<ReturnType<ResponsesParse>>>;
-interface ChatOpenAIResponsesCallOptions extends BaseChatOpenAICallOptions {
     /**
      * Configuration options for a text response from the model. Can be plain text or
      * structured JSON data.
+     *
+     * If set, the Responses API will be used to fulfill the request.
      */
-    text?: OpenAIClient.Responses.ResponseCreateParams["text"];
+    text?: ResponsesCreateParams["text"];
     /**
      * The truncation strategy to use for the model response.
+     *
+     * If set, the Responses API will be used to fulfill the request.
      */
-    truncation?: OpenAIClient.Responses.ResponseCreateParams["truncation"];
+    truncation?: ResponsesCreateParams["truncation"];
     /**
      * Specify additional output data to include in the model response.
+     *
+     * If set, the Responses API will be used to fulfill the request.
      */
-    include?: OpenAIClient.Responses.ResponseCreateParams["include"];
+    include?: ResponsesCreateParams["include"];
     /**
      * The unique ID of the previous response to the model. Use this to create multi-turn
-     * conversations.
+     * conversations. Will be set automatically if {@link ChatOpenAI.zdrEnabled} is `false`, provided
+     * that AIMessages included in the request have a `response_metadata.id` property. Ignored unless
+     * {@link ChatOpenAI.useResponsesApi} is `true`.
+     *
+     * If set, the Responses API will be used to fulfill the request.
      */
-    previous_response_id?: OpenAIClient.Responses.ResponseCreateParams["previous_response_id"];
-}
-type ChatResponsesInvocationParams = Omit<OpenAIClient.Responses.ResponseCreateParams, "input">;
-/**
- * OpenAI Responses API implementation.
- *
- * Will be exported in a later version of @langchain/openai.
- *
- * @internal
- */
-export declare class ChatOpenAIResponses<CallOptions extends ChatOpenAIResponsesCallOptions = ChatOpenAIResponsesCallOptions> extends BaseChatOpenAI<CallOptions> {
-    invocationParams(options?: this["ParsedCallOptions"]): ChatResponsesInvocationParams;
-    _generate(messages: BaseMessage[], options: this["ParsedCallOptions"]): Promise<ChatResult>;
-    _streamResponseChunks(messages: BaseMessage[], options: this["ParsedCallOptions"]): AsyncGenerator<ChatGenerationChunk>;
+    previous_response_id?: ResponsesCreateParams["previous_response_id"];
     /**
-     * Calls the Responses API with retry logic in case of failures.
-     * @param request The request to send to the OpenAI API.
-     * @param options Optional configuration for the API call.
-     * @returns The response from the OpenAI API.
+     * Service tier to use for this request. Can be "auto", "default", or "flex"
+     * Specifies the service tier for prioritization and latency optimization.
      */
-    completionWithRetry(request: OpenAIClient.Responses.ResponseCreateParamsStreaming, requestOptions?: OpenAIClient.RequestOptions): Promise<AsyncIterable<OpenAIClient.Responses.ResponseStreamEvent>>;
-    completionWithRetry(request: OpenAIClient.Responses.ResponseCreateParamsNonStreaming, requestOptions?: OpenAIClient.RequestOptions): Promise<OpenAIClient.Responses.Response>;
-    /** @internal */
-    protected _convertResponsesMessageToBaseMessage(response: ResponsesCreateInvoke | ResponsesParseInvoke): BaseMessage;
-    /** @internal */
-    protected _convertResponsesDeltaToBaseMessageChunk(chunk: OpenAIClient.Responses.ResponseStreamEvent): ChatGenerationChunk | null;
-    /** @internal */
-    protected _convertMessagesToResponsesParams(messages: BaseMessage[]): (OpenAIClient.Responses.ResponseReasoningItem | OpenAIClient.Responses.EasyInputMessage | OpenAIClient.Responses.ResponseInputItem.Message | OpenAIClient.Responses.ResponseOutputMessage | OpenAIClient.Responses.ResponseFileSearchToolCall | OpenAIClient.Responses.ResponseComputerToolCall | OpenAIClient.Responses.ResponseInputItem.ComputerCallOutput | OpenAIClient.Responses.ResponseFunctionWebSearch | OpenAIClient.Responses.ResponseFunctionToolCall | OpenAIClient.Responses.ResponseInputItem.FunctionCallOutput | OpenAIClient.Responses.ResponseInputItem.ImageGenerationCall | OpenAIClient.Responses.ResponseCodeInterpreterToolCall | OpenAIClient.Responses.ResponseInputItem.LocalShellCall | OpenAIClient.Responses.ResponseInputItem.LocalShellCallOutput | OpenAIClient.Responses.ResponseInputItem.McpListTools | OpenAIClient.Responses.ResponseInputItem.McpApprovalRequest | OpenAIClient.Responses.ResponseInputItem.McpApprovalResponse | OpenAIClient.Responses.ResponseInputItem.McpCall | OpenAIClient.Responses.ResponseInputItem.ItemReference)[];
-    /** @internal */
-    protected _convertReasoningSummary(reasoning: ChatOpenAIReasoningSummary): OpenAIClient.Responses.ResponseReasoningItem;
-    /** @internal */
-    protected _reduceChatOpenAITools(tools: ChatOpenAIToolType[], fields: {
-        stream?: boolean;
-        strict?: boolean;
-    }): ResponsesTool[];
-}
-interface ChatOpenAICompletionsCallOptions extends BaseChatOpenAICallOptions {
+    service_tier?: ResponsesCreateParams["service_tier"];
 }
-type ChatCompletionsInvocationParams = Omit<OpenAIClient.Chat.Completions.ChatCompletionCreateParams, "messages">;
-/**
- * OpenAI Completions API implementation.
- * @internal
- */
-export declare class ChatOpenAICompletions<CallOptions extends ChatOpenAICompletionsCallOptions = ChatOpenAICompletionsCallOptions> extends BaseChatOpenAI<CallOptions> {
-    /** @internal */
-    invocationParams(options?: this["ParsedCallOptions"], extra?: {
-        streaming?: boolean;
-    }): ChatCompletionsInvocationParams;
-    _generate(messages: BaseMessage[], options: this["ParsedCallOptions"], runManager?: CallbackManagerForLLMRun): Promise<ChatResult>;
-    _streamResponseChunks(messages: BaseMessage[], options: this["ParsedCallOptions"], runManager?: CallbackManagerForLLMRun): AsyncGenerator<ChatGenerationChunk>;
-    completionWithRetry(request: OpenAIClient.Chat.ChatCompletionCreateParamsStreaming, requestOptions?: OpenAIClient.RequestOptions): Promise<AsyncIterable<OpenAIClient.Chat.Completions.ChatCompletionChunk>>;
-    completionWithRetry(request: OpenAIClient.Chat.ChatCompletionCreateParamsNonStreaming, requestOptions?: OpenAIClient.RequestOptions): Promise<OpenAIClient.Chat.Completions.ChatCompletion>;
-    /** @internal */
-    protected _convertCompletionsMessageToBaseMessage(message: OpenAIClient.Chat.Completions.ChatCompletionMessage, rawResponse: OpenAIClient.Chat.Completions.ChatCompletion): BaseMessage;
-    /** @internal */
-    protected _convertCompletionsDeltaToBaseMessageChunk(delta: Record<string, any>, rawResponse: OpenAIClient.Chat.Completions.ChatCompletionChunk, defaultRole?: OpenAIClient.Chat.ChatCompletionRole): AIMessageChunk | HumanMessageChunk | SystemMessageChunk | FunctionMessageChunk | ToolMessageChunk | ChatMessageChunk;
-}
-export type ChatOpenAICallOptions = ChatOpenAICompletionsCallOptions & ChatOpenAIResponsesCallOptions;
-export interface ChatOpenAIFields extends BaseChatOpenAIFields {
+export interface ChatOpenAIFields extends Partial<OpenAIChatInput>, BaseChatModelParams {
+    configuration?: ClientOptions;
     /**
      * Whether to use the responses API for all requests. If `false` the responses API will be used
      * only when required in order to fulfill the request.
@@ -359,7 +217,7 @@ export interface ChatOpenAIFields extends BaseChatOpenAIFields {
  * import { ChatOpenAI } from '@langchain/openai';
  *
  * const llm = new ChatOpenAI({
- *   model: "gpt-4o-mini",
+ *   model: "gpt-4o",
  *   temperature: 0,
  *   maxTokens: undefined,
  *   timeout: undefined,
@@ -660,7 +518,7 @@ export interface ChatOpenAIFields extends BaseChatOpenAIFields {
  * <summary><strong>Logprobs</strong></summary>
  *
  * ```typescript
- * const logprobsLlm = new ChatOpenAI({ model: "gpt-4o-mini", logprobs: true });
+ * const logprobsLlm = new ChatOpenAI({ logprobs: true });
  * const aiMsgForLogprobs = await logprobsLlm.invoke(input);
  * console.log(aiMsgForLogprobs.response_metadata.logprobs);
  * ```
@@ -852,19 +710,159 @@ export interface ChatOpenAIFields extends BaseChatOpenAIFields {
  *
  * <br />
  */
-export declare class ChatOpenAI<CallOptions extends ChatOpenAICallOptions = ChatOpenAICallOptions> extends BaseChatOpenAI<CallOptions> implements Partial<OpenAIChatInput> {
+export declare class ChatOpenAI<CallOptions extends ChatOpenAICallOptions = ChatOpenAICallOptions> extends BaseChatModel<CallOptions, AIMessageChunk> implements Partial<OpenAIChatInput> {
+    static lc_name(): string;
+    get callKeys(): string[];
+    lc_serializable: boolean;
+    get lc_secrets(): {
+        [key: string]: string;
+    } | undefined;
+    get lc_aliases(): Record<string, string>;
+    get lc_serializable_keys(): string[];
+    temperature?: number;
+    topP?: number;
+    frequencyPenalty?: number;
+    presencePenalty?: number;
+    n?: number;
+    logitBias?: Record<string, number>;
+    /** @deprecated Use "model" instead */
+    modelName: string;
+    model: string;
+    modelKwargs?: OpenAIChatInput["modelKwargs"];
+    stop?: string[];
+    stopSequences?: string[];
+    user?: string;
+    timeout?: number;
+    streaming: boolean;
+    streamUsage: boolean;
+    maxTokens?: number;
+    logprobs?: boolean;
+    topLogprobs?: number;
+    openAIApiKey?: string;
+    apiKey?: string;
+    organization?: string;
+    __includeRawResponse?: boolean;
+    protected client: OpenAIClient;
+    protected clientConfig: ClientOptions;
+    /**
+     * Whether the model supports the `strict` argument when passing in tools.
+     * If `undefined` the `strict` argument will not be passed to OpenAI.
+     */
+    supportsStrictToolCalling?: boolean;
+    audio?: OpenAIClient.Chat.ChatCompletionAudioParam;
+    modalities?: Array<OpenAIClient.Chat.ChatCompletionModality>;
+    /**
+     * @deprecated Use {@link reasoning} object instead.
+     */
+    reasoningEffort?: OpenAIClient.Chat.ChatCompletionReasoningEffort;
+    /**
+     * Options for reasoning models.
+     */
+    reasoning?: OpenAIClient.Reasoning;
     /**
      * Whether to use the responses API for all requests. If `false` the responses API will be used
      * only when required in order to fulfill the request.
      */
     useResponsesApi: boolean;
-    private responses;
-    private completions;
-    get lc_serializable_keys(): string[];
+    /**
+     * Must be set to `true` in tenancies with Zero Data Retention. Setting to `true` will disable
+     * output storage in the Responses API, but this DOES NOT enable Zero Data Retention in your
+     * OpenAI organization or project. This must be configured directly with OpenAI.
+     *
+     * See:
+     * https://help.openai.com/en/articles/10503543-data-residency-for-the-openai-api
+     * https://platform.openai.com/docs/api-reference/responses/create#responses-create-store
+     *
+     * @default false
+     */
+    zdrEnabled?: boolean | undefined;
+    /**
+     * Service tier to use for this request. Can be "auto", "default", or "flex" or "priority".
+     * Specifies the service tier for prioritization and latency optimization.
+     */
+    service_tier?: ResponsesCreateParams["service_tier"];
     constructor(fields?: ChatOpenAIFields);
-    protected _useResponsesApi(options: this["ParsedCallOptions"] | undefined): boolean;
+    getLsParams(options: this["ParsedCallOptions"]): LangSmithParams;
+    bindTools(tools: ChatOpenAIToolType[], kwargs?: Partial<CallOptions>): Runnable<BaseLanguageModelInput, AIMessageChunk, CallOptions>;
+    private createResponseFormat;
+    private getReasoningParams;
+    /**
+     * Get the parameters used to invoke the model
+     */
+    invocationParams<Type extends "responses" | "completion" = "completion">(options?: this["ParsedCallOptions"], extra?: {
+        streaming?: boolean;
+    }): Type extends "responses" ? ResponseInvocationParams : ChatCompletionInvocationParams;
+    protected _convertOpenAIChatCompletionMessageToBaseMessage(message: OpenAIClient.Chat.Completions.ChatCompletionMessage, rawResponse: OpenAIClient.Chat.Completions.ChatCompletion): BaseMessage;
+    protected _convertOpenAIDeltaToBaseMessageChunk(delta: Record<string, any>, rawResponse: OpenAIClient.Chat.Completions.ChatCompletionChunk, defaultRole?: OpenAIRoleEnum): AIMessageChunk | HumanMessageChunk | SystemMessageChunk | FunctionMessageChunk | ToolMessageChunk | ChatMessageChunk;
     /** @ignore */
-    _generate(messages: BaseMessage[], options: this["ParsedCallOptions"], runManager?: CallbackManagerForLLMRun): Promise<ChatResult>;
+    _identifyingParams(): Omit<OpenAIClient.Chat.ChatCompletionCreateParams, "messages"> & {
+        model_name: string;
+    } & ClientOptions;
     _streamResponseChunks(messages: BaseMessage[], options: this["ParsedCallOptions"], runManager?: CallbackManagerForLLMRun): AsyncGenerator<ChatGenerationChunk>;
-    withConfig(config: Partial<CallOptions>): Runnable<BaseLanguageModelInput, AIMessageChunk, CallOptions>;
+    /**
+     * Get the identifying parameters for the model
+     *
+     */
+    identifyingParams(): Omit<OpenAIClient.Chat.Completions.ChatCompletionCreateParams, "messages"> & {
+        model_name: string;
+    } & ClientOptions;
+    /** @ignore */
+    _responseApiGenerate(messages: BaseMessage[], options: this["ParsedCallOptions"], runManager?: CallbackManagerForLLMRun): Promise<ChatResult>;
+    /**
+     * Determines whether the responses API should be used for the given request.
+     *
+     * @internal
+     *
+     * @param options The parsed call options for the request.
+     * @returns `true` if the responses API should be used, either because it is explicitly enabled,
+     * or because the request requires it.
+     */
+    protected _useResponseApi(options: this["ParsedCallOptions"] | undefined): boolean;
+    /** @ignore */
+    _generate(messages: BaseMessage[], options: this["ParsedCallOptions"], runManager?: CallbackManagerForLLMRun): Promise<ChatResult>;
+    /**
+     * Estimate the number of tokens a prompt will use.
+     * Modified from: https://github.com/hmarr/openai-chat-tokens/blob/main/src/index.ts
+     */
+    private getEstimatedTokenCountFromPrompt;
+    /**
+     * Estimate the number of tokens an array of generations have used.
+     */
+    private getNumTokensFromGenerations;
+    getNumTokensFromMessages(messages: BaseMessage[]): Promise<{
+        totalCount: number;
+        countPerMessage: number[];
+    }>;
+    /**
+     * Calls the OpenAI API with retry logic in case of failures.
+     * @param request The request to send to the OpenAI API.
+     * @param options Optional configuration for the API call.
+     * @returns The response from the OpenAI API.
+     */
+    completionWithRetry(request: OpenAIClient.Chat.ChatCompletionCreateParamsStreaming, options?: OpenAICoreRequestOptions): Promise<AsyncIterable<OpenAIClient.Chat.Completions.ChatCompletionChunk>>;
+    completionWithRetry(request: OpenAIClient.Chat.ChatCompletionCreateParamsNonStreaming, options?: OpenAICoreRequestOptions): Promise<OpenAIClient.Chat.Completions.ChatCompletion>;
+    protected responseApiWithRetry(request: ResponsesCreateParams & {
+        stream: true;
+    }, options?: OpenAICoreRequestOptions): Promise<ResponsesCreateStream>;
+    protected responseApiWithRetry(request: ResponsesCreateParams, options?: OpenAICoreRequestOptions): Promise<ResponsesCreateInvoke>;
+    /**
+     * Call the beta chat completions parse endpoint. This should only be called if
+     * response_format is set to "json_object".
+     * @param {OpenAIClient.Chat.ChatCompletionCreateParamsNonStreaming} request
+     * @param {OpenAICoreRequestOptions | undefined} options
+     */
+    betaParsedCompletionWithRetry(request: OpenAIClient.Chat.ChatCompletionCreateParamsNonStreaming, options?: OpenAICoreRequestOptions): Promise<ReturnType<OpenAIClient["chat"]["completions"]["parse"]>>;
+    protected _getClientOptions(options: OpenAICoreRequestOptions | undefined): OpenAICoreRequestOptions;
+    _llmType(): string;
+    /** @ignore */
+    _combineLLMOutput(...llmOutputs: OpenAILLMOutput[]): OpenAILLMOutput;
+    withStructuredOutput<RunOutput extends Record<string, any> = Record<string, any>>(outputSchema: InteropZodType<RunOutput> | Record<string, any>, config?: ChatOpenAIStructuredOutputMethodOptions<false>): Runnable<BaseLanguageModelInput, RunOutput>;
+    withStructuredOutput<RunOutput extends Record<string, any> = Record<string, any>>(outputSchema: InteropZodType<RunOutput> | Record<string, any>, config?: ChatOpenAIStructuredOutputMethodOptions<true>): Runnable<BaseLanguageModelInput, {
+        raw: BaseMessage;
+        parsed: RunOutput;
+    }>;
+    withStructuredOutput<RunOutput extends Record<string, any> = Record<string, any>>(outputSchema: InteropZodType<RunOutput> | Record<string, any>, config?: ChatOpenAIStructuredOutputMethodOptions<boolean>): Runnable<BaseLanguageModelInput, RunOutput> | Runnable<BaseLanguageModelInput, {
+        raw: BaseMessage;
+        parsed: RunOutput;
+    }>;
 }