ai 7.0.49 → 7.0.50

package/CHANGELOG.md

@@ -1,5 +1,24 @@
 # ai
 
+## 7.0.50
+
+### Patch Changes
+
+- 79e133c: async APIs for generateVideo (poll, webhook)
+
+  Adds an asynchronous start/status flow to the experimental video model
+  interface (`VideoModelV4`): models may now implement `doStart`, `doStatus`,
+  and `handleWebhookOption` instead of (or in addition to) `doGenerate`, and
+  `experimental_generateVideo` accepts `poll` and `webhook` options to
+  orchestrate completion via polling or webhooks. Polling configuration can use
+  a custom delay implementation for durable workflow compatibility.
+
+- da64b51: feat(code-mode): simplify tool caller configuration
+- Updated dependencies [79e133c]
+  - @ai-sdk/provider@4.0.5
+  - @ai-sdk/gateway@4.0.39
+  - @ai-sdk/provider-utils@5.0.19
+
 ## 7.0.49
 
 ### Patch Changes

package/dist/index.d.ts

@@ -1,4 +1,4 @@
-import { EmbeddingModelV4, EmbeddingModelV3, EmbeddingModelV2, EmbeddingModelV4Embedding, EmbeddingModelV4Middleware, ImageModelV4, ImageModelV3, ImageModelV2, ImageModelV4ProviderMetadata, ImageModelV2ProviderMetadata, ImageModelV4Middleware, JSONValue as JSONValue$1, LanguageModelV4, LanguageModelV3, LanguageModelV2, SharedV4Warning, LanguageModelV4Source, LanguageModelV4Middleware, RerankingModelV4, RerankingModelV3, SharedV4ProviderMetadata, SharedV4ProviderReference, SpeechModelV4, SpeechModelV3, SpeechModelV2, TranscriptionModelV4, TranscriptionModelV3, TranscriptionModelV2, JSONObject, ImageModelV4Usage, LanguageModelV4CallOptions, LanguageModelV4Prompt, AISDKError, LanguageModelV4ToolCall, JSONSchema7, ProviderV4, ProviderV3, ProviderV2, SharedV4Headers, JSONParseError, TypeValidationError, Experimental_VideoModelV4, Experimental_VideoModelV3, Experimental_VideoModelV4FrameType, EmbeddingModelV4CallOptions, Experimental_RealtimeFactoryV4, Experimental_RealtimeFactoryV4GetTokenOptions, Experimental_RealtimeFactoryV4GetTokenResult, Experimental_RealtimeModelV4, Experimental_RealtimeModelV4ClientEvent, Experimental_RealtimeModelV4ServerEvent, Experimental_RealtimeModelV4SessionConfig, Experimental_RealtimeModelV4ToolDefinition, FilesV4, SkillsV4, NoSuchModelError, SharedV4AudioFormat, Experimental_SpeechTranslationModelV4, Experimental_SpeechTranslationModelV4Usage, FilesV4UploadFileCallOptions, SkillsV4UploadSkillResult, SkillsV4UploadSkillCallOptions, SkillsV4File } from '@ai-sdk/provider';
+import { EmbeddingModelV4, EmbeddingModelV3, EmbeddingModelV2, EmbeddingModelV4Embedding, EmbeddingModelV4Middleware, ImageModelV4, ImageModelV3, ImageModelV2, ImageModelV4ProviderMetadata, ImageModelV2ProviderMetadata, ImageModelV4Middleware, JSONValue as JSONValue$1, LanguageModelV4, LanguageModelV3, LanguageModelV2, SharedV4Warning, LanguageModelV4Source, LanguageModelV4Middleware, RerankingModelV4, RerankingModelV3, SharedV4ProviderMetadata, SharedV4ProviderReference, SpeechModelV4, SpeechModelV3, SpeechModelV2, TranscriptionModelV4, TranscriptionModelV3, TranscriptionModelV2, JSONObject, ImageModelV4Usage, LanguageModelV4CallOptions, LanguageModelV4Prompt, AISDKError, LanguageModelV4ToolCall, JSONSchema7, ProviderV4, ProviderV3, ProviderV2, SharedV4Headers, JSONParseError, TypeValidationError, Experimental_VideoModelV4, Experimental_VideoModelV3, Experimental_VideoModelV4FrameType, Experimental_VideoModelV4OperationWebhook, EmbeddingModelV4CallOptions, Experimental_RealtimeFactoryV4, Experimental_RealtimeFactoryV4GetTokenOptions, Experimental_RealtimeFactoryV4GetTokenResult, Experimental_RealtimeModelV4, Experimental_RealtimeModelV4ClientEvent, Experimental_RealtimeModelV4ServerEvent, Experimental_RealtimeModelV4SessionConfig, Experimental_RealtimeModelV4ToolDefinition, FilesV4, SkillsV4, NoSuchModelError, SharedV4AudioFormat, Experimental_SpeechTranslationModelV4, Experimental_SpeechTranslationModelV4Usage, FilesV4UploadFileCallOptions, SkillsV4UploadSkillResult, SkillsV4UploadSkillCallOptions, SkillsV4File } from '@ai-sdk/provider';
 export { AISDKError, APICallError, EmptyResponseBodyError, InvalidPromptError, InvalidResponseDataError, JSONParseError, JSONSchema7, LoadAPIKeyError, LoadSettingError, NoContentGeneratedError, NoSuchModelError, NoSuchProviderReferenceError, TooManyEmbeddingValuesForCallError, TypeValidationError, UnsupportedFunctionalityError } from '@ai-sdk/provider';
 import { GatewayModelId } from '@ai-sdk/gateway';
 export { GatewayModelId, createGateway, gateway } from '@ai-sdk/gateway';
@@ -3019,19 +3019,11 @@ type ToolApprovalConfiguration<TOOLS extends ToolSet, RUNTIME_CONTEXT extends Co
     [key in keyof TOOLS]?: ToolApprovalStatus | SingleToolApprovalFunction<InferToolInput<TOOLS[key]>, InferToolContext<TOOLS[key]>, RUNTIME_CONTEXT>;
 };
 
-interface Experimental_ToolCallerReference<NAME extends string = string> {
-    readonly toolName: NAME;
-}
 type ToolCallerName<TOOLS extends ToolSet> = {
     [NAME in keyof TOOLS]: TOOLS[NAME] extends Experimental_ToolCallerTool ? NAME : never;
 }[keyof TOOLS] & string;
-type ToolCallerReferenceUnion<TOOLS extends ToolSet> = {
-    [NAME in ToolCallerName<TOOLS>]: Experimental_ToolCallerReference<NAME>;
-}[ToolCallerName<TOOLS>];
-type Experimental_ToolCallers<TOOLS extends ToolSet> = (callers: {
-    [NAME in ToolCallerName<TOOLS>]: Experimental_ToolCallerReference<NAME>;
-}) => {
-    [NAME in keyof TOOLS]?: ReadonlyArray<'direct' | ToolCallerReferenceUnion<TOOLS>>;
+type Experimental_ToolCallers<TOOLS extends ToolSet> = {
+    [NAME in keyof TOOLS]?: ReadonlyArray<'AI_SDK_DIRECT_TOOL_CALL' | ToolCallerName<TOOLS>>;
 };
 
 declare const symbol$l: unique symbol;
@@ -7881,7 +7873,48 @@ type GenerateVideoPrompt = string | {
     image: DataContent;
     text?: string;
 };
-declare function experimental_generateVideo({ model: modelArg, prompt: promptArg, n, maxVideosPerCall, aspectRatio, resolution, duration, fps, seed, frameImages, inputReferences, generateAudio, providerOptions, maxRetries: maxRetriesArg, abortSignal, headers, download: downloadFn, }: {
+/**
+ * Polling configuration for models that support the asynchronous
+ * start/status flow.
+ *
+ * When used with `webhook`, `timeoutMs` also limits how long the SDK waits for
+ * the webhook notification. If the model does not support webhooks, these
+ * options configure the automatic polling fallback.
+ */
+type GenerateVideoPollOptions = {
+    /**
+     * Interval between status checks in milliseconds.
+     *
+     * @default 5000
+     */
+    intervalMs?: number;
+    /**
+     * Maximum time to wait for completion in milliseconds.
+     *
+     * @default 600000 (10 minutes)
+     */
+    timeoutMs?: number;
+    /**
+     * Custom delay implementation for polling intervals and webhook timeouts.
+     * This can be used with durable workflow sleep functions.
+     *
+     * @default the built-in timer-based delay
+     */
+    delay?: (delayInMs: number, options?: {
+        abortSignal?: AbortSignal;
+    }) => PromiseLike<void>;
+};
+/**
+ * Webhook factory for models that support the asynchronous start/status flow.
+ *
+ * The factory should return a URL for the provider to send notifications to,
+ * and a `received` promise that resolves when the notification arrives.
+ */
+type GenerateVideoWebhookFactory = () => PromiseLike<{
+    url: string;
+    received: PromiseLike<Experimental_VideoModelV4OperationWebhook>;
+}>;
+declare function experimental_generateVideo({ model: modelArg, prompt: promptArg, n, maxVideosPerCall, aspectRatio, resolution, duration, fps, seed, frameImages, inputReferences, generateAudio, providerOptions, maxRetries: maxRetriesArg, abortSignal, headers, download: downloadFn, poll, webhook, }: {
     /**
      * The video model to use.
      */
@@ -7903,7 +7936,7 @@ declare function experimental_generateVideo({ model: modelArg, prompt: promptArg
      */
     aspectRatio?: `${number}:${number}`;
     /**
-     * Resolution of the videos to generate. Must have the format `{width}x{height}`.
+     * Resolution of the videos to generate. Must have the format `{width}x${height}`.
      */
     resolution?: `${number}x${number}`;
     /**
@@ -7985,6 +8018,27 @@ declare function experimental_generateVideo({ model: modelArg, prompt: promptArg
         data: Uint8Array;
         mediaType: string | undefined;
     }>;
+    /**
+     * Polling configuration for models that support the asynchronous
+     * start/status flow. When provided and the model implements `doStart`
+     * and `doStatus`, the SDK will orchestrate polling automatically.
+     *
+     * This option can be combined with `webhook`: `timeoutMs` limits the webhook
+     * wait, and the polling settings apply if the model does not support
+     * webhooks.
+     */
+    poll?: GenerateVideoPollOptions;
+    /**
+     * Webhook factory for models that support the asynchronous
+     * start/status flow. When provided and the model implements `doStart`
+     * and `doStatus`, the SDK will use webhooks instead of polling.
+     *
+     * The factory should return a URL for the provider to send notifications to,
+     * and a `received` promise that resolves when the notification arrives.
+     * `poll` can also be provided to configure the webhook timeout and polling
+     * fallback.
+     */
+    webhook?: GenerateVideoWebhookFactory;
 }): Promise<GenerateVideoResult>;
 
 /**
@@ -9125,4 +9179,4 @@ declare function uploadSkill({ api, files, displayTitle, providerOptions, }: {
     files: UploadSkillFile[];
 }): Promise<UploadSkillResult>;
 
-export { AI_SDK_TELEMETRY_TRACING_CHANNEL, AbstractChat, ActiveTools, Agent, AgentCallParameters, AgentStreamParameters, AsyncIterableStream, CallSettings, CallWarning, ChatAddToolApproveResponseFunction, ChatAddToolOutputFunction, ChatInit, ChatOnDataCallback, ChatOnErrorCallback, ChatOnFinishCallback, ChatOnToolCallCallback, ChatRequestOptions, ChatState, ChatStatus, ChatTransport, ChunkDetector, CompletionRequestOptions, ContentPart, CreateUIMessage, CustomContentUIPart, DataUIPart, DeepPartial, DefaultChatTransport, DefaultGeneratedFile, DirectChatTransport, DirectChatTransportOptions, DynamicToolCall, DynamicToolError, DynamicToolResult, DynamicToolUIPart, EmbedEndEvent, EmbedManyResult, EmbedResult, EmbedStartEvent, Embedding, EmbeddingModel, EmbeddingModelCallEndEvent, EmbeddingModelCallStartEvent, EmbeddingModelMiddleware, EmbeddingModelUsage, ErrorHandler, AbstractRealtimeSession as Experimental_AbstractRealtimeSession, ToolLoopAgent as Experimental_Agent, ToolLoopAgentSettings as Experimental_AgentSettings, DownloadFunction as Experimental_DownloadFunction, Experimental_GeneratedImage, InferAgentUIMessage as Experimental_InferAgentUIMessage, LanguageModelStreamPart as Experimental_LanguageModelStreamPart, LogWarningsFunction as Experimental_LogWarningsFunction, RealtimeClientEvent as Experimental_RealtimeClientEvent, RealtimeFactory as Experimental_RealtimeFactory, RealtimeFactoryGetTokenOptions as Experimental_RealtimeFactoryGetTokenOptions, RealtimeFactoryGetTokenResult as Experimental_RealtimeFactoryGetTokenResult, RealtimeModel as Experimental_RealtimeModel, RealtimeServerEvent as Experimental_RealtimeServerEvent, RealtimeSessionConfig as Experimental_RealtimeSessionConfig, RealtimeSessionOptions as Experimental_RealtimeSessionOptions, RealtimeSetupResponse as Experimental_RealtimeSetupResponse, RealtimeState as Experimental_RealtimeState, RealtimeStatus as Experimental_RealtimeStatus, RealtimeToolDefinition as Experimental_RealtimeToolDefinition, Experimental_SpeechResult, StreamTranslationResult as Experimental_StreamTranslationResult, Experimental_ToolCallerReference, Experimental_ToolCallers, Experimental_TranscriptionResult, TranslationStreamPart as Experimental_TranslationStreamPart, FileUIPart, FinishReason, GenerateImageResult, GenerateObjectEndEvent, GenerateObjectResult, GenerateObjectStartEvent, GenerateObjectStepEndEvent, GenerateObjectStepStartEvent, GenerateTextAbortEvent, GenerateTextEndEvent, GenerateTextInclude, GenerateTextOnAbortCallback, GenerateTextOnEndCallback, GenerateTextOnFinishCallback, GenerateTextOnStartCallback, GenerateTextOnStepEndCallback, GenerateTextOnStepFinishCallback, GenerateTextOnStepStartCallback, GenerateTextResult, GenerateTextStartEvent, GenerateTextStepEndEvent, GenerateTextStepStartEvent, GenerateVideoPrompt, GenerateVideoResult, GeneratedAudioFile, GeneratedFile, GenericToolApprovalFunction, HttpChatTransport, HttpChatTransportInitOptions, ImageModel, ImageModelMiddleware, ImageModelProviderMetadata, ImageModelResponseMetadata, ImageModelUsage, InferAgentUIMessage, InferCompleteOutput as InferGenerateOutput, InferPartialOutput as InferStreamOutput, InferTelemetryEvent, InferUIDataParts, InferUIMessageChunk, InferUITool, InferUITools, Instructions, InvalidArgumentError, InvalidDataContentError, InvalidMessageRoleError, InvalidStreamPartError, InvalidToolApprovalError, InvalidToolApprovalSignatureError, InvalidToolInputError, JSONValue, JsonToSseTransformStream, LanguageModel, LanguageModelCallEndEvent, LanguageModelCallOptions, LanguageModelCallStartEvent, LanguageModelMiddleware, LanguageModelRequestMetadata, LanguageModelResponseMetadata, LanguageModelUsage, LogWarningsFunction, MessageConversionError, MissingToolResultsError, ModelInfo, NoImageGeneratedError, NoObjectGeneratedError, NoOutputGeneratedError, NoSpeechGeneratedError, NoSuchProviderError, NoSuchToolError, NoTranscriptGeneratedError, NoTranslationGeneratedError, NoVideoGeneratedError, ObjectStreamPart, OnFinishEvent, OnLanguageModelCallEndCallback, OnLanguageModelCallStartCallback, OnStartEvent, OnStepFinishEvent, OnStepStartEvent, OnToolCallFinishEvent, OnToolCallStartEvent, OnToolExecutionEndCallback, OnToolExecutionStartCallback, output as Output, OutputChunkTimingStats, PrepareReconnectToStreamRequest, PrepareSendMessagesRequest, PrepareStepFunction, PrepareStepResult, Prompt, Provider, ProviderMetadata, ProviderReference, ProviderRegistryProvider, ReasoningFileOutput, ReasoningFileUIPart, ReasoningOutput, ReasoningUIPart, RepairTextFunction, RequestOptions, RerankEndEvent, RerankResult, RerankStartEvent, RerankingModel, RerankingModelCallEndEvent, RerankingModelCallStartEvent, RetryError, SafeValidateUIMessagesResult, SerialJobExecutor, SingleToolApprovalFunction, SourceDocumentUIPart, SourceUrlUIPart, SpeechModel, SpeechModelResponseMetadata, SpeechResult, StaticToolCall, StaticToolError, StaticToolOutputDenied, StaticToolResult, StepResult, StepResultPerformance, StepStartUIPart, StopCondition, StreamObjectOnFinishCallback, StreamObjectResult, StreamTextInclude, StreamTextOnChunkCallback, StreamTextOnErrorCallback, StreamTextResult, StreamTextTransform, StreamTranscriptionResult, Telemetry, TelemetryOptions, TelemetryTracingChannelMessage, TelemetryTracingEventType, TextStreamChatTransport, TextStreamPart, TextUIPart, TimeoutConfiguration, ToUIMessageChunkOptions, ToolApprovalConfiguration, ToolApprovalRequestOutput, ToolApprovalResponseOutput, ToolApprovalStatus, ToolCallNotFoundForApprovalError, ToolCallRepairError, ToolCallRepairFunction, ToolChoice, ToolExecutionEndEvent, ToolExecutionStartEvent, ToolInputRefinement, ToolLoopAgent, ToolLoopAgentSettings, ToolOrder, ToolUIPart, TranscriptionModel, TranscriptionModelResponseMetadata, TranscriptionResult, TranscriptionStreamPart, TypedToolCall, TypedToolError, TypedToolOutputDenied, TypedToolResult, UIDataPartSchemas, UIDataTypes, UIMessage, UIMessageChunk, UIMessagePart, UIMessageStreamError, UIMessageStreamOnEndCallback, UIMessageStreamOnFinishCallback, UIMessageStreamOnStepEndCallback, UIMessageStreamOnStepFinishCallback, UIMessageStreamOptions, UIMessageStreamWriter, UITool, UIToolInvocation, UITools, UI_MESSAGE_STREAM_HEADERS, UnsupportedModelVersionError, UploadFileResult, UploadSkillResult, UseCompletionOptions, Warning, addToolInputExamplesMiddleware, assistantModelMessageSchema, callCompletionApi, consumeStream, convertDataContentToBase64String, convertFileListToFileUIParts, convertToModelMessages, cosineSimilarity, createAgentUIStream, createAgentUIStreamResponse, createDownload, createProviderRegistry, createTextStreamResponse, createUIMessageStream, createUIMessageStreamResponse, customProvider, defaultEmbeddingSettingsMiddleware, defaultSettingsMiddleware, detectToolDrift, embed, embedMany, experimental_createProviderRegistry, decodeRealtimeAudio as experimental_decodeRealtimeAudio, encodeRealtimeAudio as experimental_encodeRealtimeAudio, filterActiveTools as experimental_filterActiveTools, experimental_generateSpeech, experimental_generateVideo, getRealtimeToolDefinitions as experimental_getRealtimeToolDefinitions, resampleAudio as experimental_resampleAudio, streamLanguageModelCall as experimental_streamLanguageModelCall, streamTranscribe as experimental_streamTranscribe, streamTranslate as experimental_streamTranslate, experimental_transcribe, extractJsonMiddleware, extractReasoningMiddleware, fingerprintTools, generateImage, generateObject, generateSpeech, generateText, getChunkTimeoutMs, getFirstChunkTimeoutMs, getStaticToolName, getStepTimeoutMs, getTextFromDataUrl, getToolName, getToolOrDynamicToolName, getToolTimeoutMs, getTotalTimeoutMs, hasToolCall, isCustomContentUIPart, isDataUIPart, isDeepEqualData, isDynamicToolUIPart, isFileUIPart, isLoopFinished, isReasoningFileUIPart, isReasoningUIPart, isStaticToolUIPart, isStepCount, isTextUIPart, isToolUIPart, lastAssistantMessageIsCompleteWithApprovalResponses, lastAssistantMessageIsCompleteWithToolCalls, modelMessageSchema, parsePartialJson, pipeAgentUIStreamToResponse, pipeTextStreamToResponse, pipeUIMessageStreamToResponse, pruneMessages, readUIMessageStream, registerTelemetry, rerank, safeValidateUIMessages, simulateReadableStream, simulateStreamingMiddleware, smoothStream, isStepCount as stepCountIs, streamObject, streamText, systemModelMessageSchema, toTextStream, toUIMessageChunk, toUIMessageStream, toolModelMessageSchema, transcribe, uiMessageChunkSchema, uploadFile, uploadSkill, userModelMessageSchema, validateUIMessages, wrapEmbeddingModel, wrapImageModel, wrapLanguageModel, wrapProvider };
+export { AI_SDK_TELEMETRY_TRACING_CHANNEL, AbstractChat, ActiveTools, Agent, AgentCallParameters, AgentStreamParameters, AsyncIterableStream, CallSettings, CallWarning, ChatAddToolApproveResponseFunction, ChatAddToolOutputFunction, ChatInit, ChatOnDataCallback, ChatOnErrorCallback, ChatOnFinishCallback, ChatOnToolCallCallback, ChatRequestOptions, ChatState, ChatStatus, ChatTransport, ChunkDetector, CompletionRequestOptions, ContentPart, CreateUIMessage, CustomContentUIPart, DataUIPart, DeepPartial, DefaultChatTransport, DefaultGeneratedFile, DirectChatTransport, DirectChatTransportOptions, DynamicToolCall, DynamicToolError, DynamicToolResult, DynamicToolUIPart, EmbedEndEvent, EmbedManyResult, EmbedResult, EmbedStartEvent, Embedding, EmbeddingModel, EmbeddingModelCallEndEvent, EmbeddingModelCallStartEvent, EmbeddingModelMiddleware, EmbeddingModelUsage, ErrorHandler, AbstractRealtimeSession as Experimental_AbstractRealtimeSession, ToolLoopAgent as Experimental_Agent, ToolLoopAgentSettings as Experimental_AgentSettings, DownloadFunction as Experimental_DownloadFunction, Experimental_GeneratedImage, InferAgentUIMessage as Experimental_InferAgentUIMessage, LanguageModelStreamPart as Experimental_LanguageModelStreamPart, LogWarningsFunction as Experimental_LogWarningsFunction, RealtimeClientEvent as Experimental_RealtimeClientEvent, RealtimeFactory as Experimental_RealtimeFactory, RealtimeFactoryGetTokenOptions as Experimental_RealtimeFactoryGetTokenOptions, RealtimeFactoryGetTokenResult as Experimental_RealtimeFactoryGetTokenResult, RealtimeModel as Experimental_RealtimeModel, RealtimeServerEvent as Experimental_RealtimeServerEvent, RealtimeSessionConfig as Experimental_RealtimeSessionConfig, RealtimeSessionOptions as Experimental_RealtimeSessionOptions, RealtimeSetupResponse as Experimental_RealtimeSetupResponse, RealtimeState as Experimental_RealtimeState, RealtimeStatus as Experimental_RealtimeStatus, RealtimeToolDefinition as Experimental_RealtimeToolDefinition, Experimental_SpeechResult, StreamTranslationResult as Experimental_StreamTranslationResult, Experimental_ToolCallers, Experimental_TranscriptionResult, TranslationStreamPart as Experimental_TranslationStreamPart, FileUIPart, FinishReason, GenerateImageResult, GenerateObjectEndEvent, GenerateObjectResult, GenerateObjectStartEvent, GenerateObjectStepEndEvent, GenerateObjectStepStartEvent, GenerateTextAbortEvent, GenerateTextEndEvent, GenerateTextInclude, GenerateTextOnAbortCallback, GenerateTextOnEndCallback, GenerateTextOnFinishCallback, GenerateTextOnStartCallback, GenerateTextOnStepEndCallback, GenerateTextOnStepFinishCallback, GenerateTextOnStepStartCallback, GenerateTextResult, GenerateTextStartEvent, GenerateTextStepEndEvent, GenerateTextStepStartEvent, GenerateVideoPrompt, GenerateVideoResult, GeneratedAudioFile, GeneratedFile, GenericToolApprovalFunction, HttpChatTransport, HttpChatTransportInitOptions, ImageModel, ImageModelMiddleware, ImageModelProviderMetadata, ImageModelResponseMetadata, ImageModelUsage, InferAgentUIMessage, InferCompleteOutput as InferGenerateOutput, InferPartialOutput as InferStreamOutput, InferTelemetryEvent, InferUIDataParts, InferUIMessageChunk, InferUITool, InferUITools, Instructions, InvalidArgumentError, InvalidDataContentError, InvalidMessageRoleError, InvalidStreamPartError, InvalidToolApprovalError, InvalidToolApprovalSignatureError, InvalidToolInputError, JSONValue, JsonToSseTransformStream, LanguageModel, LanguageModelCallEndEvent, LanguageModelCallOptions, LanguageModelCallStartEvent, LanguageModelMiddleware, LanguageModelRequestMetadata, LanguageModelResponseMetadata, LanguageModelUsage, LogWarningsFunction, MessageConversionError, MissingToolResultsError, ModelInfo, NoImageGeneratedError, NoObjectGeneratedError, NoOutputGeneratedError, NoSpeechGeneratedError, NoSuchProviderError, NoSuchToolError, NoTranscriptGeneratedError, NoTranslationGeneratedError, NoVideoGeneratedError, ObjectStreamPart, OnFinishEvent, OnLanguageModelCallEndCallback, OnLanguageModelCallStartCallback, OnStartEvent, OnStepFinishEvent, OnStepStartEvent, OnToolCallFinishEvent, OnToolCallStartEvent, OnToolExecutionEndCallback, OnToolExecutionStartCallback, output as Output, OutputChunkTimingStats, PrepareReconnectToStreamRequest, PrepareSendMessagesRequest, PrepareStepFunction, PrepareStepResult, Prompt, Provider, ProviderMetadata, ProviderReference, ProviderRegistryProvider, ReasoningFileOutput, ReasoningFileUIPart, ReasoningOutput, ReasoningUIPart, RepairTextFunction, RequestOptions, RerankEndEvent, RerankResult, RerankStartEvent, RerankingModel, RerankingModelCallEndEvent, RerankingModelCallStartEvent, RetryError, SafeValidateUIMessagesResult, SerialJobExecutor, SingleToolApprovalFunction, SourceDocumentUIPart, SourceUrlUIPart, SpeechModel, SpeechModelResponseMetadata, SpeechResult, StaticToolCall, StaticToolError, StaticToolOutputDenied, StaticToolResult, StepResult, StepResultPerformance, StepStartUIPart, StopCondition, StreamObjectOnFinishCallback, StreamObjectResult, StreamTextInclude, StreamTextOnChunkCallback, StreamTextOnErrorCallback, StreamTextResult, StreamTextTransform, StreamTranscriptionResult, Telemetry, TelemetryOptions, TelemetryTracingChannelMessage, TelemetryTracingEventType, TextStreamChatTransport, TextStreamPart, TextUIPart, TimeoutConfiguration, ToUIMessageChunkOptions, ToolApprovalConfiguration, ToolApprovalRequestOutput, ToolApprovalResponseOutput, ToolApprovalStatus, ToolCallNotFoundForApprovalError, ToolCallRepairError, ToolCallRepairFunction, ToolChoice, ToolExecutionEndEvent, ToolExecutionStartEvent, ToolInputRefinement, ToolLoopAgent, ToolLoopAgentSettings, ToolOrder, ToolUIPart, TranscriptionModel, TranscriptionModelResponseMetadata, TranscriptionResult, TranscriptionStreamPart, TypedToolCall, TypedToolError, TypedToolOutputDenied, TypedToolResult, UIDataPartSchemas, UIDataTypes, UIMessage, UIMessageChunk, UIMessagePart, UIMessageStreamError, UIMessageStreamOnEndCallback, UIMessageStreamOnFinishCallback, UIMessageStreamOnStepEndCallback, UIMessageStreamOnStepFinishCallback, UIMessageStreamOptions, UIMessageStreamWriter, UITool, UIToolInvocation, UITools, UI_MESSAGE_STREAM_HEADERS, UnsupportedModelVersionError, UploadFileResult, UploadSkillResult, UseCompletionOptions, Warning, addToolInputExamplesMiddleware, assistantModelMessageSchema, callCompletionApi, consumeStream, convertDataContentToBase64String, convertFileListToFileUIParts, convertToModelMessages, cosineSimilarity, createAgentUIStream, createAgentUIStreamResponse, createDownload, createProviderRegistry, createTextStreamResponse, createUIMessageStream, createUIMessageStreamResponse, customProvider, defaultEmbeddingSettingsMiddleware, defaultSettingsMiddleware, detectToolDrift, embed, embedMany, experimental_createProviderRegistry, decodeRealtimeAudio as experimental_decodeRealtimeAudio, encodeRealtimeAudio as experimental_encodeRealtimeAudio, filterActiveTools as experimental_filterActiveTools, experimental_generateSpeech, experimental_generateVideo, getRealtimeToolDefinitions as experimental_getRealtimeToolDefinitions, resampleAudio as experimental_resampleAudio, streamLanguageModelCall as experimental_streamLanguageModelCall, streamTranscribe as experimental_streamTranscribe, streamTranslate as experimental_streamTranslate, experimental_transcribe, extractJsonMiddleware, extractReasoningMiddleware, fingerprintTools, generateImage, generateObject, generateSpeech, generateText, getChunkTimeoutMs, getFirstChunkTimeoutMs, getStaticToolName, getStepTimeoutMs, getTextFromDataUrl, getToolName, getToolOrDynamicToolName, getToolTimeoutMs, getTotalTimeoutMs, hasToolCall, isCustomContentUIPart, isDataUIPart, isDeepEqualData, isDynamicToolUIPart, isFileUIPart, isLoopFinished, isReasoningFileUIPart, isReasoningUIPart, isStaticToolUIPart, isStepCount, isTextUIPart, isToolUIPart, lastAssistantMessageIsCompleteWithApprovalResponses, lastAssistantMessageIsCompleteWithToolCalls, modelMessageSchema, parsePartialJson, pipeAgentUIStreamToResponse, pipeTextStreamToResponse, pipeUIMessageStreamToResponse, pruneMessages, readUIMessageStream, registerTelemetry, rerank, safeValidateUIMessages, simulateReadableStream, simulateStreamingMiddleware, smoothStream, isStepCount as stepCountIs, streamObject, streamText, systemModelMessageSchema, toTextStream, toUIMessageChunk, toUIMessageStream, toolModelMessageSchema, transcribe, uiMessageChunkSchema, uploadFile, uploadSkill, userModelMessageSchema, validateUIMessages, wrapEmbeddingModel, wrapImageModel, wrapLanguageModel, wrapProvider };

package/dist/index.js

@@ -1143,7 +1143,7 @@ import {
 } from "@ai-sdk/provider-utils";
 
 // src/version.ts
-var VERSION = true ? "7.0.49" : "0.0.0-test";
+var VERSION = true ? "7.0.50" : "0.0.0-test";
 
 // src/util/download/download.ts
 var download = async ({
@@ -4881,6 +4881,7 @@ function sortToolResultContentByToolCallOrder({
 import {
   experimental_getToolCaller
 } from "@ai-sdk/provider-utils";
+var DIRECT_TOOL_CALL = "AI_SDK_DIRECT_TOOL_CALL";
 function resolveToolCallerConfiguration({
   tools,
   toolCallers
@@ -4888,46 +4889,34 @@ function resolveToolCallerConfiguration({
   if (tools == null || toolCallers == null) {
     return void 0;
   }
-  const namesByReference = /* @__PURE__ */ new WeakMap();
-  const callerReferences = {};
-  for (const [toolName, tool2] of Object.entries(tools)) {
-    if (experimental_getToolCaller(tool2) == null) {
-      continue;
-    }
-    const reference = Object.freeze({ toolName });
-    namesByReference.set(reference, toolName);
-    callerReferences[toolName] = reference;
-  }
-  const configuration = toolCallers(callerReferences);
   const resolved = {};
-  for (const [toolName, callers] of Object.entries(configuration)) {
+  for (const [toolName, callers] of Object.entries(toolCallers)) {
     if (!Object.prototype.hasOwnProperty.call(tools, toolName)) {
       throw new InvalidArgumentError({
         parameter: "experimental_toolCallers",
-        value: configuration,
+        value: toolCallers,
         message: `unknown tool "${toolName}".`
       });
     }
     if (!Array.isArray(callers)) {
       throw new InvalidArgumentError({
         parameter: "experimental_toolCallers",
-        value: configuration,
+        value: toolCallers,
         message: `callers for tool "${toolName}" must be an array.`
       });
     }
     resolved[toolName] = callers.map((caller) => {
-      if (caller === "direct") {
+      if (caller === DIRECT_TOOL_CALL) {
         return caller;
       }
-      const callerName = caller != null && typeof caller === "object" ? namesByReference.get(caller) : void 0;
-      if (callerName == null) {
+      if (typeof caller !== "string" || !Object.prototype.hasOwnProperty.call(tools, caller) || experimental_getToolCaller(tools[caller]) == null) {
         throw new InvalidArgumentError({
           parameter: "experimental_toolCallers",
-          value: configuration,
-          message: `tool "${toolName}" contains an invalid caller reference.`
+          value: toolCallers,
+          message: `tool "${toolName}" contains an invalid caller.`
         });
       }
-      return callerName;
+      return caller;
     });
   }
   return resolved;
@@ -4952,7 +4941,7 @@ function prepareToolsForToolCallers({
     let availableToProvider = false;
     let preparedTool = tool2;
     for (const callerName of callerNames) {
-      if (callerName === "direct") {
+      if (callerName === DIRECT_TOOL_CALL) {
         availableDirectly = true;
         continue;
       }
@@ -13895,6 +13884,7 @@ function detectToolDrift(current, baseline) {
 // src/generate-video/generate-video.ts
 import {
   convertBase64ToUint8Array as convertBase64ToUint8Array5,
+  delay as defaultDelay,
   withUserAgentSuffix as withUserAgentSuffix9,
   detectMediaType as detectMediaType4
 } from "@ai-sdk/provider-utils";
@@ -13916,7 +13906,9 @@ async function experimental_generateVideo({
   maxRetries: maxRetriesArg,
   abortSignal,
   headers,
-  download: downloadFn = defaultDownload
+  download: downloadFn = defaultDownload,
+  poll,
+  webhook
 }) {
   var _a23, _b;
   const model = resolveVideoModel(modelArg);
@@ -13956,32 +13948,59 @@ async function experimental_generateVideo({
   }
   const resolvedImage = firstFrameImage != null ? firstFrameImage : image;
   const maxVideosPerCallWithDefault = (_b = maxVideosPerCall != null ? maxVideosPerCall : await invokeModelMaxVideosPerCall(model)) != null ? _b : 1;
+  const hasStartStatus = model.doStart != null && model.doStatus != null;
+  const useStartStatus = hasStartStatus && (poll != null || webhook != null || model.doGenerate == null);
+  if (model.doGenerate == null && !hasStartStatus) {
+    throw new Error(
+      `Video model ${model.modelId} does not implement doGenerate or doStart/doStatus.`
+    );
+  }
+  if ((poll != null || webhook != null) && !hasStartStatus) {
+    logWarnings({
+      warnings: [
+        {
+          type: "other",
+          message: "poll/webhook options were provided but the model does not support doStart/doStatus. Falling back to doGenerate."
+        }
+      ],
+      provider: model.provider,
+      model: model.modelId
+    });
+  }
   const callCount = Math.ceil(n / maxVideosPerCallWithDefault);
   const callVideoCounts = Array.from({ length: callCount }, (_, index) => {
     const remaining = n - index * maxVideosPerCallWithDefault;
     return Math.min(remaining, maxVideosPerCallWithDefault);
   });
   const results = await Promise.all(
-    callVideoCounts.map(
-      async (callVideoCount) => await retry(
-        () => model.doGenerate({
-          prompt,
-          n: callVideoCount,
-          aspectRatio,
-          resolution,
-          duration,
-          fps,
-          seed,
-          image: resolvedImage,
-          frameImages: normalizedFrameImages,
-          inputReferences: effectiveInputReferences,
-          generateAudio,
-          providerOptions: providerOptions != null ? providerOptions : {},
-          headers: headersWithUserAgent,
-          abortSignal
-        })
-      )
-    )
+    callVideoCounts.map(async (callVideoCount) => {
+      const callOptions = {
+        prompt,
+        n: callVideoCount,
+        aspectRatio,
+        resolution,
+        duration,
+        fps,
+        seed,
+        image: resolvedImage,
+        frameImages: normalizedFrameImages,
+        inputReferences: effectiveInputReferences,
+        generateAudio,
+        providerOptions: providerOptions != null ? providerOptions : {},
+        headers: headersWithUserAgent,
+        abortSignal
+      };
+      if (useStartStatus) {
+        return executeStartStatusFlow({
+          model,
+          callOptions,
+          poll,
+          webhook,
+          retry
+        });
+      }
+      return retry(() => model.doGenerate(callOptions));
+    })
   );
   const videos = [];
   const responses = [];
@@ -14039,25 +14058,7 @@ async function experimental_generateVideo({
       providerMetadata: result.providerMetadata
     });
     if (result.providerMetadata != null) {
-      for (const [providerName, metadata] of Object.entries(
-        result.providerMetadata
-      )) {
-        const existingMetadata = providerMetadata[providerName];
-        if (existingMetadata != null && typeof existingMetadata === "object") {
-          providerMetadata[providerName] = {
-            ...existingMetadata,
-            ...metadata
-          };
-          if ("videos" in existingMetadata && Array.isArray(existingMetadata.videos) && "videos" in metadata && Array.isArray(metadata.videos)) {
-            providerMetadata[providerName].videos = [
-              ...existingMetadata.videos,
-              ...metadata.videos
-            ];
-          }
-        } else {
-          providerMetadata[providerName] = metadata;
-        }
-      }
+      mergeProviderMetadata(providerMetadata, result.providerMetadata);
     }
   }
   if (videos.length === 0) {
@@ -14078,6 +14079,139 @@ async function experimental_generateVideo({
     providerMetadata
   };
 }
+async function executeStartStatusFlow({
+  model,
+  callOptions,
+  poll: pollConfig,
+  webhook: webhookFactory,
+  retry
+}) {
+  var _a23, _b, _c;
+  const earlyWarnings = [];
+  let webhookUrl;
+  let webhookReceived;
+  if (webhookFactory != null) {
+    if (model.handleWebhookOption != null) {
+      const result = await model.handleWebhookOption({
+        webhook: webhookFactory
+      });
+      webhookUrl = result.webhookUrl;
+      webhookReceived = result.received;
+    } else {
+      earlyWarnings.push({
+        type: "unsupported",
+        feature: "webhook",
+        details: "This model does not support webhooks. Falling back to polling."
+      });
+    }
+  }
+  const startResult = await retry(
+    () => model.doStart({
+      ...callOptions,
+      webhookUrl
+    })
+  );
+  const allWarnings = [...earlyWarnings, ...startResult.warnings];
+  let operationProviderMetadata = startResult.providerMetadata == null ? void 0 : { ...startResult.providerMetadata };
+  const intervalMs = (_a23 = pollConfig == null ? void 0 : pollConfig.intervalMs) != null ? _a23 : 5e3;
+  const timeoutMs = (_b = pollConfig == null ? void 0 : pollConfig.timeoutMs) != null ? _b : 6e5;
+  const delay = (_c = pollConfig == null ? void 0 : pollConfig.delay) != null ? _c : defaultDelay;
+  const startTime = Date.now();
+  if (webhookReceived != null) {
+    await waitForWebhook({
+      received: webhookReceived,
+      timeoutMs,
+      abortSignal: callOptions.abortSignal,
+      delay
+    });
+  }
+  while (true) {
+    if (webhookReceived == null) {
+      const elapsedMs = Date.now() - startTime;
+      if (elapsedMs >= timeoutMs) {
+        throw new Error(`Video generation timed out after ${timeoutMs}ms.`);
+      }
+      await delay(Math.min(intervalMs, timeoutMs - elapsedMs), {
+        abortSignal: callOptions.abortSignal
+      });
+      if (Date.now() - startTime >= timeoutMs) {
+        throw new Error(`Video generation timed out after ${timeoutMs}ms.`);
+      }
+    }
+    const statusResult = await retry(
+      () => model.doStatus({
+        operation: startResult.operation,
+        abortSignal: callOptions.abortSignal,
+        headers: callOptions.headers
+      })
+    );
+    if (statusResult.status === "error") {
+      throw new Error(statusResult.error);
+    }
+    if (statusResult.warnings != null) {
+      allWarnings.push(...statusResult.warnings);
+    }
+    if (statusResult.providerMetadata != null) {
+      operationProviderMetadata != null ? operationProviderMetadata : operationProviderMetadata = {};
+      mergeProviderMetadata(
+        operationProviderMetadata,
+        statusResult.providerMetadata
+      );
+    }
+    if (statusResult.status === "completed") {
+      return {
+        videos: statusResult.videos,
+        warnings: allWarnings,
+        providerMetadata: operationProviderMetadata,
+        response: statusResult.response
+      };
+    }
+    if (webhookReceived != null) {
+      throw new Error(
+        "Video generation did not complete after webhook notification."
+      );
+    }
+  }
+}
+async function waitForWebhook({
+  received,
+  timeoutMs,
+  abortSignal,
+  delay
+}) {
+  const timeoutController = typeof globalThis.AbortController === "function" ? new globalThis.AbortController() : void 0;
+  try {
+    await Promise.race([
+      received,
+      delay(timeoutMs, {
+        abortSignal: timeoutController == null ? abortSignal : mergeAbortSignals(abortSignal, timeoutController.signal)
+      }).then(() => {
+        throw new Error(`Video generation timed out after ${timeoutMs}ms.`);
+      })
+    ]);
+  } finally {
+    timeoutController == null ? void 0 : timeoutController.abort();
+  }
+}
+function mergeProviderMetadata(target, source) {
+  for (const [providerName, metadataValue] of Object.entries(source)) {
+    const existingMetadata = target[providerName];
+    if (existingMetadata != null && typeof existingMetadata === "object" && metadataValue != null && typeof metadataValue === "object") {
+      target[providerName] = {
+        ...existingMetadata,
+        ...metadataValue
+      };
+      if ("videos" in existingMetadata && Array.isArray(existingMetadata.videos) && "videos" in metadataValue && Array.isArray(metadataValue.videos)) {
+        target[providerName].videos = [
+          ...existingMetadata.videos,
+          ...metadataValue.videos
+        ];
+      }
+    } else {
+      target[providerName] = metadataValue;
+    }
+  }
+}
 function normalizePrompt2(promptArg) {
   if (typeof promptArg === "string") {
     return {