@ai-sdk/open-responses 1.0.0

package/src/responses/open-responses-api.ts

@@ -0,0 +1,1239 @@
+import { JSONSchema7 } from '@ai-sdk/provider';
+import { lazySchema } from '@ai-sdk/provider-utils';
+import { z } from 'zod/v4';
+import { zodSchema } from '@ai-sdk/provider-utils';
+
+export const openResponsesErrorSchema = lazySchema(() =>
+  zodSchema(
+    z.object({
+      error: z.object({
+        message: z.string(),
+        type: z.string(),
+        param: z.string(),
+        code: z.string(),
+      }),
+    }),
+  ),
+);
+
+// ============================================================================
+// Enums
+// ============================================================================
+
+/**
+ * The status of a function call or message item.
+ */
+export type FunctionCallStatus = 'in_progress' | 'completed' | 'incomplete';
+
+/**
+ * Image detail level for input images.
+ */
+export type ImageDetail = 'low' | 'high' | 'auto';
+
+/**
+ * Reasoning effort level.
+ */
+export type ReasoningEffortEnum = 'none' | 'low' | 'medium' | 'high' | 'xhigh';
+
+/**
+ * Reasoning summary level.
+ */
+export type ReasoningSummaryEnum = 'concise' | 'detailed' | 'auto';
+
+/**
+ * Tool choice value enum.
+ */
+export type ToolChoiceValueEnum = 'none' | 'auto' | 'required';
+
+/**
+ * Verbosity level for text output.
+ */
+export type VerbosityEnum = 'low' | 'medium' | 'high';
+
+// ============================================================================
+// Content Types
+// ============================================================================
+
+/**
+ * A text input to the model.
+ */
+export type InputTextContentParam = {
+  type: 'input_text';
+  text: string;
+};
+
+/**
+ * An image input to the model.
+ */
+export type InputImageContentParam = {
+  type: 'input_image';
+  image_url?: string;
+  detail?: ImageDetail;
+};
+
+/**
+ * A file input to the model.
+ */
+export type InputFileContentParam = {
+  type: 'input_file';
+  filename?: string;
+  file_data?: string;
+  file_url?: string;
+};
+
+/**
+ * A video input to the model.
+ */
+export type InputVideoContent = {
+  type: 'input_video';
+  video_url: string;
+};
+
+/**
+ * A text output from the model.
+ */
+export type OutputTextContentParam = {
+  type: 'output_text';
+  text: string;
+  annotations?: UrlCitationParam[];
+};
+
+/**
+ * A refusal from the model.
+ */
+export type RefusalContentParam = {
+  type: 'refusal';
+  refusal: string;
+};
+
+/**
+ * A URL citation annotation.
+ */
+export type UrlCitationParam = {
+  type: 'url_citation';
+  start_index: number;
+  end_index: number;
+  url: string;
+  title: string;
+};
+
+/**
+ * Reasoning summary content.
+ */
+export type ReasoningSummaryContentParam = {
+  type: 'summary_text';
+  text: string;
+};
+
+// ============================================================================
+// Message Item Types
+// ============================================================================
+
+/**
+ * An internal identifier for an item to reference.
+ */
+export type ItemReferenceParam = {
+  type?: 'item_reference';
+  id: string;
+};
+
+/**
+ * A reasoning item.
+ */
+export type ReasoningItemParam = {
+  id?: string;
+  type: 'reasoning';
+  summary: ReasoningSummaryContentParam[];
+  content?: unknown;
+  encrypted_content?: string;
+};
+
+/**
+ * A user message item.
+ */
+export type UserMessageItemParam = {
+  id?: string;
+  type: 'message';
+  role: 'user';
+  content:
+    | string
+    | Array<
+        InputTextContentParam | InputImageContentParam | InputFileContentParam
+      >;
+  status?: string;
+};
+
+/**
+ * A system message item.
+ */
+export type SystemMessageItemParam = {
+  id?: string;
+  type: 'message';
+  role: 'system';
+  content: string | InputTextContentParam[];
+  status?: string;
+};
+
+/**
+ * A developer message item.
+ */
+export type DeveloperMessageItemParam = {
+  id?: string;
+  type: 'message';
+  role: 'developer';
+  content: string | InputTextContentParam[];
+  status?: string;
+};
+
+/**
+ * An assistant message item.
+ */
+export type AssistantMessageItemParam = {
+  id?: string;
+  type: 'message';
+  role: 'assistant';
+  content: string | Array<OutputTextContentParam | RefusalContentParam>;
+  status?: string;
+};
+
+/**
+ * A function call item.
+ */
+export type FunctionCallItemParam = {
+  id?: string;
+  call_id: string;
+  type: 'function_call';
+  name: string;
+  arguments: string;
+  status?: FunctionCallStatus;
+};
+
+/**
+ * A function call output item.
+ */
+export type FunctionCallOutputItemParam = {
+  id?: string;
+  call_id: string;
+  type: 'function_call_output';
+  output:
+    | string
+    | Array<
+        | InputTextContentParam
+        | InputImageContentParam
+        | InputFileContentParam
+        | InputVideoContent
+      >;
+  status?: FunctionCallStatus;
+};
+
+// ============================================================================
+// Tool Types
+// ============================================================================
+
+/**
+ * A function tool parameter.
+ */
+export type FunctionToolParam = {
+  name: string;
+  description?: string;
+  parameters?: JSONSchema7;
+  strict?: boolean;
+  type: 'function';
+};
+
+/**
+ * A specific function tool choice.
+ */
+export type SpecificFunctionParam = {
+  type: 'function';
+  name: string;
+};
+
+/**
+ * Allowed tools parameter.
+ */
+export type AllowedToolsParam = {
+  type: 'allowed_tools';
+  tools: SpecificFunctionParam[];
+  mode?: ToolChoiceValueEnum;
+};
+
+/**
+ * Controls which tool the model should use, if any.
+ */
+export type ToolChoiceParam =
+  | ToolChoiceValueEnum
+  | SpecificFunctionParam
+  | AllowedToolsParam;
+
+// ============================================================================
+// Configuration Types
+// ============================================================================
+
+/**
+ * Set of 16 key-value pairs that can be attached to an object.
+ */
+export type MetadataParam = Record<string, string>;
+
+/**
+ * Text response format (plain text).
+ */
+export type TextResponseFormat = {
+  type: 'text';
+};
+
+/**
+ * JSON schema response format.
+ */
+export type JsonSchemaResponseFormatParam = {
+  type: 'json_schema';
+  description?: string;
+  name?: string;
+  schema?: JSONSchema7;
+  strict?: boolean;
+};
+
+/**
+ * Configuration options for text output.
+ */
+export type TextParam = {
+  format?: TextResponseFormat | JsonSchemaResponseFormatParam;
+  verbosity?: VerbosityEnum;
+};
+
+/**
+ * Options that control streamed response behavior.
+ */
+export type StreamOptionsParam = {
+  include_obfuscation?: boolean;
+};
+
+/**
+ * Configuration options for reasoning behavior.
+ */
+export type ReasoningParam = {
+  effort?: ReasoningEffortEnum;
+  summary?: ReasoningSummaryEnum;
+};
+
+// ============================================================================
+// Response-Specific Types
+// ============================================================================
+
+/**
+ * The status of a message item in the response.
+ */
+export type MessageStatus = 'in_progress' | 'completed' | 'incomplete';
+
+/**
+ * Truncation enum for responses.
+ */
+export type TruncationEnum = 'auto' | 'disabled';
+
+/**
+ * Service tier enum.
+ */
+export type ServiceTierEnum = 'auto' | 'default' | 'flex' | 'priority';
+
+/**
+ * A top log probability of a token.
+ */
+export type TopLogProb = {
+  token: string;
+  logprob: number;
+  bytes: number[];
+};
+
+/**
+ * The log probability of a token.
+ */
+export type LogProb = {
+  token: string;
+  logprob: number;
+  bytes: number[];
+  top_logprobs: TopLogProb[];
+};
+
+/**
+ * A URL citation annotation in a response.
+ */
+export type UrlCitationBody = {
+  type: 'url_citation';
+  url: string;
+  start_index: number;
+  end_index: number;
+  title: string;
+};
+
+/**
+ * An annotation that applies to a span of output text.
+ */
+export type Annotation = UrlCitationBody;
+
+/**
+ * A text input content in a response.
+ */
+export type InputTextContent = {
+  type: 'input_text';
+  text: string;
+};
+
+/**
+ * A text output from the model in a response.
+ */
+export type OutputTextContent = {
+  type: 'output_text';
+  text: string;
+  annotations: Annotation[];
+  logprobs: LogProb[];
+};
+
+/**
+ * A refusal from the model in a response.
+ */
+export type RefusalContent = {
+  type: 'refusal';
+  refusal: string;
+};
+
+/**
+ * Reasoning text from the model.
+ */
+export type ReasoningTextContent = {
+  type: 'reasoning_text';
+  text: string;
+};
+
+/**
+ * A summary text from the model.
+ */
+export type SummaryTextContent = {
+  type: 'summary_text';
+  text: string;
+};
+
+/**
+ * An image input content in a response.
+ */
+export type InputImageContent = {
+  type: 'input_image';
+  image_url?: string;
+  detail: ImageDetail;
+};
+
+/**
+ * A file input content in a response.
+ */
+export type InputFileContent = {
+  type: 'input_file';
+  filename?: string;
+  file_url?: string;
+};
+
+/**
+ * A message in the response.
+ */
+export type Message = {
+  type: 'message';
+  id: string;
+  status: MessageStatus;
+  role: 'user' | 'assistant' | 'system' | 'developer';
+  content: InputTextContent[];
+};
+
+/**
+ * A function tool call that was generated by the model.
+ */
+export type FunctionCall = {
+  type: 'function_call';
+  id: string;
+  call_id: string;
+  name: string;
+  arguments: string;
+  status: FunctionCallStatus;
+};
+
+/**
+ * A function tool call output that was returned.
+ */
+export type FunctionCallOutput = {
+  type: 'function_call_output';
+  id: string;
+  call_id: string;
+  output:
+    | string
+    | Array<InputTextContent | InputImageContent | InputFileContent>;
+  status: FunctionCallStatus;
+};
+
+/**
+ * A reasoning item that was generated by the model.
+ */
+export type ReasoningBody = {
+  type: 'reasoning';
+  id: string;
+  content?: InputTextContent[];
+  summary: InputTextContent[];
+  encrypted_content?: string;
+};
+
+/**
+ * Output item field union type.
+ */
+export type OutputItem =
+  | FunctionCall
+  | FunctionCallOutput
+  | Message
+  | ReasoningBody;
+
+/**
+ * Details about why the response was incomplete.
+ */
+export type IncompleteDetails = {
+  reason: string;
+};
+
+/**
+ * An error that occurred while generating the response.
+ */
+export type ResponseError = {
+  code: string;
+  message: string;
+};
+
+/**
+ * A function tool in a response.
+ */
+export type FunctionTool = {
+  type: 'function';
+  name: string;
+  description?: string;
+  parameters?: JSONSchema7;
+  strict?: boolean;
+};
+
+/**
+ * Function tool choice in a response.
+ */
+export type FunctionToolChoice = {
+  type: 'function';
+  name?: string;
+};
+
+/**
+ * Allowed tool choice in a response.
+ */
+export type AllowedToolChoice = {
+  type: 'allowed_tools';
+  tools: FunctionToolChoice[];
+  mode: ToolChoiceValueEnum;
+};
+
+/**
+ * Tool choice in a response.
+ */
+export type ResponseToolChoice =
+  | ToolChoiceValueEnum
+  | FunctionToolChoice
+  | AllowedToolChoice;
+
+/**
+ * JSON object response format.
+ */
+export type JsonObjectResponseFormat = {
+  type: 'json_object';
+};
+
+/**
+ * JSON schema response format in a response.
+ */
+export type JsonSchemaResponseFormat = {
+  type: 'json_schema';
+  name: string;
+  description?: string;
+  schema: unknown;
+  strict: boolean;
+};
+
+/**
+ * Text field in a response.
+ */
+export type TextField = {
+  format?:
+    | TextResponseFormat
+    | JsonObjectResponseFormat
+    | JsonSchemaResponseFormat;
+  verbosity?: VerbosityEnum;
+};
+
+/**
+ * A breakdown of input token usage that was recorded.
+ */
+export type InputTokensDetails = {
+  cached_tokens: number;
+};
+
+/**
+ * A breakdown of output token usage that was recorded.
+ */
+export type OutputTokensDetails = {
+  reasoning_tokens: number;
+};
+
+/**
+ * Token usage statistics that were recorded for the response.
+ */
+export type Usage = {
+  input_tokens: number;
+  output_tokens: number;
+  total_tokens: number;
+  input_tokens_details: InputTokensDetails;
+  output_tokens_details: OutputTokensDetails;
+};
+
+/**
+ * Reasoning configuration and outputs that were produced for this response.
+ */
+export type Reasoning = {
+  effort?: ReasoningEffortEnum;
+  summary?: ReasoningSummaryEnum;
+};
+
+// ============================================================================
+// Request Body
+// ============================================================================
+
+/**
+ * Body that is sent to the Open Responses API.
+ */
+export type OpenResponsesRequestBody = {
+  /**
+   * The model to use for this request, e.g. 'gpt-5.2'.
+   */
+  model: string;
+
+  /**
+   * Context for the model: either a string (interpreted as a user message),
+   * or an array of structured message items.
+   */
+  input:
+    | string
+    | Array<
+        | ItemReferenceParam
+        | ReasoningItemParam
+        | UserMessageItemParam
+        | SystemMessageItemParam
+        | DeveloperMessageItemParam
+        | AssistantMessageItemParam
+        | FunctionCallItemParam
+        | FunctionCallOutputItemParam
+      >;
+
+  /**
+   * The ID of the response to use as the prior turn for this request.
+   */
+  previous_response_id?: string;
+
+  /**
+   * Options specifying extra values to include in the response.
+   */
+  include?: Array<
+    'reasoning.encrypted_content' | 'message.output_text.logprobs'
+  >;
+
+  /**
+   * A list of tools that the model may call while generating the response.
+   */
+  tools?: FunctionToolParam[];
+
+  /**
+   * Controls which tool the model should use, if any.
+   */
+  tool_choice?: ToolChoiceParam;
+
+  /**
+   * Structured metadata as up to 16 key-value pairs.
+   */
+  metadata?: MetadataParam;
+
+  /**
+   * Configuration options for text output.
+   */
+  text?: TextParam;
+
+  /**
+   * Sampling temperature to use, between 0 and 2.
+   */
+  temperature?: number;
+
+  /**
+   * Nucleus sampling parameter, between 0 and 1.
+   */
+  top_p?: number;
+
+  /**
+   * Penalizes new tokens based on whether they appear in the text so far.
+   */
+  presence_penalty?: number;
+
+  /**
+   * Penalizes new tokens based on their frequency in the text so far.
+   */
+  frequency_penalty?: number;
+
+  /**
+   * Whether the model may call multiple tools in parallel.
+   */
+  parallel_tool_calls?: boolean;
+
+  /**
+   * Whether to stream response events as server-sent events.
+   */
+  stream?: boolean;
+
+  /**
+   * Options that control streamed response behavior.
+   */
+  stream_options?: StreamOptionsParam;
+
+  /**
+   * Whether to run the request in the background and return immediately.
+   */
+  background?: boolean;
+
+  /**
+   * Maximum number of tokens the model may generate.
+   */
+  max_output_tokens?: number;
+
+  /**
+   * Maximum number of tool calls the model may make while generating the response.
+   */
+  max_tool_calls?: number;
+
+  /**
+   * Configuration options for reasoning behavior.
+   */
+  reasoning?: ReasoningParam;
+
+  /**
+   * A stable identifier used for safety monitoring and abuse detection.
+   */
+  safety_identifier?: string;
+
+  /**
+   * A key to use when reading/writing to the prompt cache.
+   */
+  prompt_cache_key?: string;
+
+  /**
+   * Controls how input is truncated if it exceeds the model's context window.
+   * - 'auto': Let the service decide how to truncate.
+   * - 'disabled': Disable truncation. Context overflow yields 400 error.
+   */
+  truncation?: 'auto' | 'disabled';
+
+  /**
+   * Additional instructions to guide the model for this request.
+   */
+  instructions?: string;
+
+  /**
+   * Whether to store the response so it can be retrieved later.
+   */
+  store?: boolean;
+
+  /**
+   * The service tier to use for this request.
+   * - 'auto' | 'default' | 'flex' | 'priority'
+   */
+  service_tier?: 'auto' | 'default' | 'flex' | 'priority';
+
+  /**
+   * Number of most likely tokens to return at each position, with logprobs.
+   */
+  top_logprobs?: number;
+};
+
+// ============================================================================
+// Response Body
+// ============================================================================
+
+/**
+ * Response body from the Open Responses API.
+ */
+export type OpenResponsesResponseBody = {
+  /**
+   * The unique ID of the response that was created.
+   */
+  id: string;
+
+  /**
+   * The object type, which is always 'response'.
+   */
+  object: 'response';
+
+  /**
+   * The Unix timestamp (in seconds) for when the response was created.
+   */
+  created_at: number;
+
+  /**
+   * The Unix timestamp (in seconds) for when the response was completed, if it was completed.
+   */
+  completed_at?: number;
+
+  /**
+   * The status that was set for the response.
+   */
+  status: string;
+
+  /**
+   * Details about why the response was incomplete, if applicable.
+   */
+  incomplete_details?: IncompleteDetails;
+
+  /**
+   * The model that generated this response.
+   */
+  model: string;
+
+  /**
+   * The ID of the previous response in the chain that was referenced, if any.
+   */
+  previous_response_id?: string;
+
+  /**
+   * Additional instructions that were used to guide the model for this response.
+   */
+  instructions?: string;
+
+  /**
+   * The output items that were generated by the model.
+   */
+  output: OutputItem[];
+
+  /**
+   * The error that occurred, if the response failed.
+   */
+  error?: ResponseError;
+
+  /**
+   * The tools that were available to the model during response generation.
+   */
+  tools?: FunctionTool[];
+
+  /**
+   * The tool choice configuration that was used.
+   */
+  tool_choice?: ResponseToolChoice;
+
+  /**
+   * How the input was truncated by the service when it exceeded the model context window.
+   */
+  truncation?: TruncationEnum;
+
+  /**
+   * Whether the model was allowed to call multiple tools in parallel.
+   */
+  parallel_tool_calls?: boolean;
+
+  /**
+   * Configuration options for text output that were used.
+   */
+  text?: TextField;
+
+  /**
+   * The nucleus sampling parameter that was used for this response.
+   */
+  top_p?: number;
+
+  /**
+   * The presence penalty that was used.
+   */
+  presence_penalty?: number;
+
+  /**
+   * The frequency penalty that was used.
+   */
+  frequency_penalty?: number;
+
+  /**
+   * The number of most likely tokens that were returned at each position.
+   */
+  top_logprobs?: number;
+
+  /**
+   * The sampling temperature that was used for this response.
+   */
+  temperature?: number;
+
+  /**
+   * Reasoning configuration and outputs that were produced for this response.
+   */
+  reasoning?: Reasoning;
+
+  /**
+   * Token usage statistics that were recorded for the response, if available.
+   */
+  usage?: Usage;
+
+  /**
+   * The maximum number of tokens the model was allowed to generate for this response.
+   */
+  max_output_tokens?: number;
+
+  /**
+   * The maximum number of tool calls the model was allowed to make.
+   */
+  max_tool_calls?: number;
+
+  /**
+   * Whether this response was stored so it can be retrieved later.
+   */
+  store?: boolean;
+
+  /**
+   * Whether this request was run in the background.
+   */
+  background?: boolean;
+
+  /**
+   * The service tier that was used for this response.
+   */
+  service_tier?: string;
+
+  /**
+   * Developer-defined metadata that was associated with the response.
+   */
+  metadata?: unknown;
+
+  /**
+   * A stable identifier that was used for safety monitoring and abuse detection.
+   */
+  safety_identifier?: string;
+
+  /**
+   * A key that was used to read from or write to the prompt cache.
+   */
+  prompt_cache_key?: string;
+};
+
+// ============================================================================
+// Streaming Chunk Types
+// ============================================================================
+
+/**
+ * Content part for streaming - output text.
+ */
+export type OutputTextContentPart = {
+  type: 'output_text';
+  text: string;
+  annotations: Annotation[];
+};
+
+/**
+ * Content part for streaming - refusal.
+ */
+export type RefusalContentPart = {
+  type: 'refusal';
+  refusal: string;
+};
+
+/**
+ * Union of content parts that can appear in streaming.
+ */
+export type ContentPart = OutputTextContentPart | RefusalContentPart;
+
+// ----------------------------------------------------------------------------
+// State Machine Events
+// ----------------------------------------------------------------------------
+
+/**
+ * Emitted when a response is created.
+ */
+export type ResponseCreatedEvent = {
+  type: 'response.created';
+  sequence_number: number;
+  response: OpenResponsesResponseBody;
+};
+
+/**
+ * Emitted when a response transitions to in_progress status.
+ */
+export type ResponseInProgressEvent = {
+  type: 'response.in_progress';
+  sequence_number: number;
+  response: OpenResponsesResponseBody;
+};
+
+/**
+ * Emitted when a response completes successfully.
+ */
+export type ResponseCompletedEvent = {
+  type: 'response.completed';
+  sequence_number: number;
+  response: OpenResponsesResponseBody;
+};
+
+/**
+ * Emitted when a response fails.
+ */
+export type ResponseFailedEvent = {
+  type: 'response.failed';
+  sequence_number: number;
+  response: OpenResponsesResponseBody;
+};
+
+/**
+ * Emitted when a response is incomplete (e.g., token budget exhausted).
+ */
+export type ResponseIncompleteEvent = {
+  type: 'response.incomplete';
+  sequence_number: number;
+  response: OpenResponsesResponseBody;
+};
+
+// ----------------------------------------------------------------------------
+// Delta Events - Output Items
+// ----------------------------------------------------------------------------
+
+/**
+ * Emitted when a new output item is added to the response.
+ */
+export type ResponseOutputItemAddedEvent = {
+  type: 'response.output_item.added';
+  sequence_number: number;
+  output_index: number;
+  item: OutputItem;
+};
+
+/**
+ * Emitted when an output item is completed.
+ */
+export type ResponseOutputItemDoneEvent = {
+  type: 'response.output_item.done';
+  sequence_number: number;
+  output_index: number;
+  item: OutputItem;
+};
+
+// ----------------------------------------------------------------------------
+// Delta Events - Content Parts
+// ----------------------------------------------------------------------------
+
+/**
+ * Emitted when a new content part is added to an item.
+ */
+export type ResponseContentPartAddedEvent = {
+  type: 'response.content_part.added';
+  sequence_number: number;
+  item_id: string;
+  output_index: number;
+  content_index: number;
+  part: ContentPart;
+};
+
+/**
+ * Emitted when a content part is completed.
+ */
+export type ResponseContentPartDoneEvent = {
+  type: 'response.content_part.done';
+  sequence_number: number;
+  item_id: string;
+  output_index: number;
+  content_index: number;
+  part: ContentPart;
+};
+
+// ----------------------------------------------------------------------------
+// Delta Events - Text Output
+// ----------------------------------------------------------------------------
+
+/**
+ * Emitted when text is appended to an output.
+ */
+export type ResponseOutputTextDeltaEvent = {
+  type: 'response.output_text.delta';
+  sequence_number: number;
+  item_id: string;
+  output_index: number;
+  content_index: number;
+  delta: string;
+  logprobs?: LogProb[];
+};
+
+/**
+ * Emitted when text output is complete.
+ */
+export type ResponseOutputTextDoneEvent = {
+  type: 'response.output_text.done';
+  sequence_number: number;
+  item_id: string;
+  output_index: number;
+  content_index: number;
+  text: string;
+  logprobs?: LogProb[];
+};
+
+// ----------------------------------------------------------------------------
+// Delta Events - Refusal
+// ----------------------------------------------------------------------------
+
+/**
+ * Emitted when refusal text is appended.
+ */
+export type ResponseRefusalDeltaEvent = {
+  type: 'response.refusal.delta';
+  sequence_number: number;
+  item_id: string;
+  output_index: number;
+  content_index: number;
+  delta: string;
+};
+
+/**
+ * Emitted when refusal is complete.
+ */
+export type ResponseRefusalDoneEvent = {
+  type: 'response.refusal.done';
+  sequence_number: number;
+  item_id: string;
+  output_index: number;
+  content_index: number;
+  refusal: string;
+};
+
+// ----------------------------------------------------------------------------
+// Delta Events - Function Call Arguments
+// ----------------------------------------------------------------------------
+
+/**
+ * Emitted when function call arguments are appended.
+ */
+export type ResponseFunctionCallArgumentsDeltaEvent = {
+  type: 'response.function_call_arguments.delta';
+  sequence_number: number;
+  item_id: string;
+  output_index: number;
+  call_id: string;
+  delta: string;
+};
+
+/**
+ * Emitted when function call arguments are complete.
+ */
+export type ResponseFunctionCallArgumentsDoneEvent = {
+  type: 'response.function_call_arguments.done';
+  sequence_number: number;
+  item_id: string;
+  output_index: number;
+  call_id: string;
+  arguments: string;
+};
+
+// ----------------------------------------------------------------------------
+// Delta Events - Reasoning
+// ----------------------------------------------------------------------------
+
+/**
+ * Emitted when reasoning summary text is appended.
+ */
+export type ResponseReasoningSummaryTextDeltaEvent = {
+  type: 'response.reasoning_summary_text.delta';
+  sequence_number: number;
+  item_id: string;
+  output_index: number;
+  summary_index: number;
+  delta: string;
+};
+
+/**
+ * Emitted when reasoning summary text is complete.
+ */
+export type ResponseReasoningSummaryTextDoneEvent = {
+  type: 'response.reasoning_summary_text.done';
+  sequence_number: number;
+  item_id: string;
+  output_index: number;
+  summary_index: number;
+  text: string;
+};
+
+/**
+ * Emitted when a reasoning summary part is added.
+ */
+export type ResponseReasoningSummaryPartAddedEvent = {
+  type: 'response.reasoning_summary_part.added';
+  sequence_number: number;
+  item_id: string;
+  output_index: number;
+  summary_index: number;
+  part: SummaryTextContent;
+};
+
+/**
+ * Emitted when a reasoning summary part is complete.
+ */
+export type ResponseReasoningSummaryPartDoneEvent = {
+  type: 'response.reasoning_summary_part.done';
+  sequence_number: number;
+  item_id: string;
+  output_index: number;
+  summary_index: number;
+  part: SummaryTextContent;
+};
+
+// ----------------------------------------------------------------------------
+// Error Event
+// ----------------------------------------------------------------------------
+
+/**
+ * Emitted when an error occurs during streaming.
+ */
+export type ResponseErrorEvent = {
+  type: 'error';
+  sequence_number: number;
+  error: ResponseError;
+};
+
+// ----------------------------------------------------------------------------
+// Union Type for All Streaming Chunks
+// ----------------------------------------------------------------------------
+
+/**
+ * Union of all streaming chunk event types from the Open Responses API.
+ *
+ * Streaming events fall into two categories:
+ * - **State Machine Events**: Represent status changes (e.g., `response.in_progress`, `response.completed`)
+ * - **Delta Events**: Represent incremental changes (e.g., `response.output_text.delta`, `response.output_item.added`)
+ */
+export type OpenResponsesChunk =
+  // State Machine Events
+  | ResponseCreatedEvent
+  | ResponseInProgressEvent
+  | ResponseCompletedEvent
+  | ResponseFailedEvent
+  | ResponseIncompleteEvent
+  // Delta Events - Output Items
+  | ResponseOutputItemAddedEvent
+  | ResponseOutputItemDoneEvent
+  // Delta Events - Content Parts
+  | ResponseContentPartAddedEvent
+  | ResponseContentPartDoneEvent
+  // Delta Events - Text Output
+  | ResponseOutputTextDeltaEvent
+  | ResponseOutputTextDoneEvent
+  // Delta Events - Refusal
+  | ResponseRefusalDeltaEvent
+  | ResponseRefusalDoneEvent
+  // Delta Events - Function Call Arguments
+  | ResponseFunctionCallArgumentsDeltaEvent
+  | ResponseFunctionCallArgumentsDoneEvent
+  // Delta Events - Reasoning
+  | ResponseReasoningSummaryTextDeltaEvent
+  | ResponseReasoningSummaryTextDoneEvent
+  | ResponseReasoningSummaryPartAddedEvent
+  | ResponseReasoningSummaryPartDoneEvent
+  // Error Event
+  | ResponseErrorEvent;