@google/genai 2.1.0 → 2.3.0

package/README.md

@@ -369,7 +369,10 @@ async function main() {
 main();
 ```
 
-## Interactions
+## Interactions (Experimental)
+
+> **Warning:** The Interactions API is in **Beta**. This is a preview of an
+experimental feature. Features and schemas are subject to **breaking changes**.
 
 The Interactions API is a unified interface for interacting with Gemini models
 and agents. It simplifies state management, tool orchestration, and long-running

package/dist/genai.d.ts

@@ -700,7 +700,7 @@ declare interface BaseCreateAgentInteractionParams {
      * Body param: Enforces that the generated response is a JSON object that complies
      * with the JSON schema specified in this field.
      */
-    response_format?: Array<AudioResponseFormat | TextResponseFormat | ImageResponseFormat | VideoResponseFormat | unknown> | AudioResponseFormat | TextResponseFormat | ImageResponseFormat | VideoResponseFormat | unknown;
+    response_format?: Array<AudioResponseFormat | TextResponseFormat | ImageResponseFormat | unknown> | AudioResponseFormat | TextResponseFormat | ImageResponseFormat | unknown;
     /**
      * Body param: The mime type of the response. This is required if response_format
      * is set.
@@ -767,7 +767,7 @@ declare interface BaseCreateModelInteractionParams {
      * Body param: Enforces that the generated response is a JSON object that complies
      * with the JSON schema specified in this field.
      */
-    response_format?: Array<AudioResponseFormat | TextResponseFormat | ImageResponseFormat | VideoResponseFormat | unknown> | AudioResponseFormat | TextResponseFormat | ImageResponseFormat | VideoResponseFormat | unknown;
+    response_format?: Array<AudioResponseFormat | TextResponseFormat | ImageResponseFormat | unknown> | AudioResponseFormat | TextResponseFormat | ImageResponseFormat | unknown;
     /**
      * Body param: The mime type of the response. This is required if response_format
      * is set.
@@ -6280,7 +6280,7 @@ declare interface Interaction {
      * Enforces that the generated response is a JSON object that complies with the
      * JSON schema specified in this field.
      */
-    response_format?: Array<AudioResponseFormat | TextResponseFormat | ImageResponseFormat | VideoResponseFormat | unknown> | AudioResponseFormat | TextResponseFormat | ImageResponseFormat | VideoResponseFormat | unknown;
+    response_format?: Array<AudioResponseFormat | TextResponseFormat | ImageResponseFormat | unknown> | AudioResponseFormat | TextResponseFormat | ImageResponseFormat | unknown;
     /**
      * The mime type of the response. This is required if response_format is set.
      */
@@ -6314,6 +6314,30 @@ declare interface Interaction {
      * completes.
      */
     webhook_config?: WebhookConfig_2;
+    /**
+     * Concatenated text from the last model output in response to the current request.
+     *
+     * Note: this is added by the SDK.
+     */
+    output_text?: string;
+    /**
+     * The last image generated by the model in response to the current request.
+     *
+     * Note: this is added by the SDK.
+     */
+    output_image?: ImageContent;
+    /**
+     * The last audio generated by the model in response to the current request.
+     *
+     * Note: this is added by the SDK.
+     */
+    output_audio?: AudioContent;
+    /**
+     * The last video generated by the model in response to the current request.
+     *
+     * Note: this is added by the SDK.
+     */
+    output_video?: VideoContent;
 }
 
 declare interface InteractionCancelParams {
@@ -6408,7 +6432,7 @@ export declare class Interactions extends BaseInteractions {
 }
 
 export declare namespace Interactions {
-    export { type AllowedTools as AllowedTools, type Annotation as Annotation, type AudioContent as AudioContent, type AudioResponseFormat as AudioResponseFormat, type CodeExecutionCallArguments as CodeExecutionCallArguments, type CodeExecutionCallStep as CodeExecutionCallStep, type CodeExecutionResultStep as CodeExecutionResultStep, type Content_2 as Content, type DeepResearchAgentConfig as DeepResearchAgentConfig, type DocumentContent as DocumentContent, type DynamicAgentConfig as DynamicAgentConfig, type ErrorEvent_2 as ErrorEvent, type FileCitation as FileCitation, type FileSearchCallStep as FileSearchCallStep, type FileSearchResultStep as FileSearchResultStep, type Function_2 as Function, type FunctionCallStep as FunctionCallStep, type FunctionResultStep as FunctionResultStep, type GenerationConfig_2 as GenerationConfig, type GoogleMapsCallArguments as GoogleMapsCallArguments, type GoogleMapsCallStep as GoogleMapsCallStep, type GoogleMapsResult as GoogleMapsResult, type GoogleMapsResultStep as GoogleMapsResultStep, type GoogleSearchCallArguments as GoogleSearchCallArguments, type GoogleSearchCallStep as GoogleSearchCallStep, type GoogleSearchResult as GoogleSearchResult, type GoogleSearchResultStep as GoogleSearchResultStep, type ImageConfig_2 as ImageConfig, type ImageContent as ImageContent, type ImageResponseFormat as ImageResponseFormat, type Interaction as Interaction, type InteractionCompletedEvent as InteractionCompletedEvent, type InteractionCreatedEvent as InteractionCreatedEvent, type InteractionSSEEvent as InteractionSSEEvent, type InteractionStatusUpdate as InteractionStatusUpdate, type MCPServerToolCallStep as MCPServerToolCallStep, type MCPServerToolResultStep as MCPServerToolResultStep, type Model_2 as Model, type ModelOutputStep as ModelOutputStep, type PlaceCitation as PlaceCitation, type SpeechConfig_2 as SpeechConfig, type Step as Step, type StepDelta as StepDelta, type StepStart as StepStart, type StepStop as StepStop, type TextContent as TextContent, type TextResponseFormat as TextResponseFormat, type ThinkingLevel_2 as ThinkingLevel, type ThoughtStep as ThoughtStep, type Tool_2 as Tool, type ToolChoiceConfig as ToolChoiceConfig, type ToolChoiceType as ToolChoiceType, type URLCitation as URLCitation, type URLContextCallArguments as URLContextCallArguments, type URLContextCallStep as URLContextCallStep, type URLContextResult as URLContextResult, type URLContextResultStep as URLContextResultStep, type Usage as Usage, type UserInputStep as UserInputStep, type VideoContent as VideoContent, type VideoResponseFormat as VideoResponseFormat, type WebhookConfig_2 as WebhookConfig, type InteractionDeleteResponse as InteractionDeleteResponse, type InteractionCreateParams as InteractionCreateParams, type CreateModelInteractionParamsNonStreaming as CreateModelInteractionParamsNonStreaming, type CreateModelInteractionParamsStreaming as CreateModelInteractionParamsStreaming, type CreateAgentInteractionParamsNonStreaming as CreateAgentInteractionParamsNonStreaming, type CreateAgentInteractionParamsStreaming as CreateAgentInteractionParamsStreaming, type InteractionDeleteParams as InteractionDeleteParams, type InteractionCancelParams as InteractionCancelParams, type InteractionGetParams as InteractionGetParams, type InteractionGetParamsNonStreaming as InteractionGetParamsNonStreaming, type InteractionGetParamsStreaming as InteractionGetParamsStreaming, };
+    export { type AllowedTools as AllowedTools, type Annotation as Annotation, type AudioContent as AudioContent, type AudioResponseFormat as AudioResponseFormat, type CodeExecutionCallArguments as CodeExecutionCallArguments, type CodeExecutionCallStep as CodeExecutionCallStep, type CodeExecutionResultStep as CodeExecutionResultStep, type Content_2 as Content, type DeepResearchAgentConfig as DeepResearchAgentConfig, type DocumentContent as DocumentContent, type DynamicAgentConfig as DynamicAgentConfig, type ErrorEvent_2 as ErrorEvent, type FileCitation as FileCitation, type FileSearchCallStep as FileSearchCallStep, type FileSearchResultStep as FileSearchResultStep, type Function_2 as Function, type FunctionCallStep as FunctionCallStep, type FunctionResultStep as FunctionResultStep, type GenerationConfig_2 as GenerationConfig, type GoogleMapsCallArguments as GoogleMapsCallArguments, type GoogleMapsCallStep as GoogleMapsCallStep, type GoogleMapsResult as GoogleMapsResult, type GoogleMapsResultStep as GoogleMapsResultStep, type GoogleSearchCallArguments as GoogleSearchCallArguments, type GoogleSearchCallStep as GoogleSearchCallStep, type GoogleSearchResult as GoogleSearchResult, type GoogleSearchResultStep as GoogleSearchResultStep, type ImageConfig_2 as ImageConfig, type ImageContent as ImageContent, type ImageResponseFormat as ImageResponseFormat, type Interaction as Interaction, type InteractionCompletedEvent as InteractionCompletedEvent, type InteractionCreatedEvent as InteractionCreatedEvent, type InteractionSSEEvent as InteractionSSEEvent, type InteractionStatusUpdate as InteractionStatusUpdate, type MCPServerToolCallStep as MCPServerToolCallStep, type MCPServerToolResultStep as MCPServerToolResultStep, type Model_2 as Model, type ModelOutputStep as ModelOutputStep, type PlaceCitation as PlaceCitation, type SpeechConfig_2 as SpeechConfig, type Step as Step, type StepDelta as StepDelta, type StepStart as StepStart, type StepStop as StepStop, type TextContent as TextContent, type TextResponseFormat as TextResponseFormat, type ThinkingLevel_2 as ThinkingLevel, type ThoughtStep as ThoughtStep, type Tool_2 as Tool, type ToolChoiceConfig as ToolChoiceConfig, type ToolChoiceType as ToolChoiceType, type URLCitation as URLCitation, type URLContextCallArguments as URLContextCallArguments, type URLContextCallStep as URLContextCallStep, type URLContextResult as URLContextResult, type URLContextResultStep as URLContextResultStep, type Usage as Usage, type UserInputStep as UserInputStep, type VideoContent as VideoContent, type WebhookConfig_2 as WebhookConfig, type InteractionDeleteResponse as InteractionDeleteResponse, type InteractionCreateParams as InteractionCreateParams, type CreateModelInteractionParamsNonStreaming as CreateModelInteractionParamsNonStreaming, type CreateModelInteractionParamsStreaming as CreateModelInteractionParamsStreaming, type CreateAgentInteractionParamsNonStreaming as CreateAgentInteractionParamsNonStreaming, type CreateAgentInteractionParamsStreaming as CreateAgentInteractionParamsStreaming, type InteractionDeleteParams as InteractionDeleteParams, type InteractionCancelParams as InteractionCancelParams, type InteractionGetParams as InteractionGetParams, type InteractionGetParamsNonStreaming as InteractionGetParamsNonStreaming, type InteractionGetParamsStreaming as InteractionGetParamsStreaming, };
 }
 
 declare namespace InteractionsAPI {
@@ -6475,7 +6499,6 @@ declare namespace InteractionsAPI {
         Usage,
         UserInputStep,
         VideoContent,
-        VideoResponseFormat,
         WebhookConfig_2 as WebhookConfig,
         InteractionDeleteResponse,
         InteractionCreateParams,
@@ -7136,6 +7159,8 @@ export declare interface LiveConnectConfig {
      response.
      */
     safetySettings?: SafetySetting[];
+    /** Config for stream translation. */
+    streamTranslationConfig?: StreamTranslationConfig;
 }
 
 /** Config for LiveConnectConstraints for Auth Token creation. */
@@ -9962,7 +9987,7 @@ export declare enum StartSensitivity {
 declare type Step = UserInputStep | ModelOutputStep | ThoughtStep | FunctionCallStep | CodeExecutionCallStep | URLContextCallStep | MCPServerToolCallStep | GoogleSearchCallStep | FileSearchCallStep | GoogleMapsCallStep | FunctionResultStep | CodeExecutionResultStep | URLContextResultStep | GoogleSearchResultStep | MCPServerToolResultStep | FileSearchResultStep | GoogleMapsResultStep;
 
 declare interface StepDelta {
-    delta: StepDelta.Text | StepDelta.Image | StepDelta.Audio | StepDelta.Document | StepDelta.Video | StepDelta.ThoughtSummary | StepDelta.ThoughtSignature | StepDelta.TextAnnotationDelta | StepDelta.ArgumentsDelta | StepDelta.CodeExecutionCall | StepDelta.URLContextCall | StepDelta.GoogleSearchCall | StepDelta.MCPServerToolCall | StepDelta.FileSearchCall | StepDelta.GoogleMapsCall | StepDelta.CodeExecutionResult | StepDelta.URLContextResult | StepDelta.GoogleSearchResult | StepDelta.MCPServerToolResult | StepDelta.FileSearchResult | StepDelta.GoogleMapsResult;
+    delta: StepDelta.Text | StepDelta.Image | StepDelta.Audio | StepDelta.Document | StepDelta.Video | StepDelta.ThoughtSummary | StepDelta.ThoughtSignature | StepDelta.TextAnnotationDelta | StepDelta.ArgumentsDelta | StepDelta.CodeExecutionCall | StepDelta.URLContextCall | StepDelta.GoogleSearchCall | StepDelta.MCPServerToolCall | StepDelta.FileSearchCall | StepDelta.GoogleMapsCall | StepDelta.CodeExecutionResult | StepDelta.URLContextResult | StepDelta.GoogleSearchResult | StepDelta.MCPServerToolResult | StepDelta.FileSearchResult | StepDelta.GoogleMapsResult | StepDelta.FunctionResult;
     event_type: 'step.delta';
     index: number;
     /**
@@ -10043,6 +10068,7 @@ declare namespace StepDelta {
     }
     interface ArgumentsDelta {
         type: 'arguments_delta';
+        arguments?: string;
     }
     interface CodeExecutionCall {
         /**
@@ -10163,6 +10189,16 @@ declare namespace StepDelta {
          */
         signature?: string;
     }
+    interface FunctionResult {
+        /**
+         * Required. ID to match the ID from the function call block.
+         */
+        call_id: string;
+        result: unknown | Array<InteractionsAPI.TextContent | InteractionsAPI.ImageContent> | string;
+        type: 'function_result';
+        is_error?: boolean;
+        name?: string;
+    }
 }
 
 declare interface StepStart {
@@ -10226,6 +10262,17 @@ export declare interface StreamableHttpTransport {
     url?: string;
 }
 
+/** Config for stream translation. */
+export declare interface StreamTranslationConfig {
+    /** If true, the model will generate audio when the target language is
+     spoken, essentially it will parrot the input. If false, we will not produce
+     audio for the target language. */
+    echoTargetLanguage?: boolean;
+    /** The target language for translation. Supported values are BCP-47
+     language codes (e.g. "en", "es", "fr"). */
+    targetLanguageCode?: string;
+}
+
 /** User provided string values assigned to a single metadata key. This data type is not supported in Vertex AI. */
 export declare interface StringList {
     /** The string values of the metadata to store. */
@@ -11823,6 +11870,7 @@ declare namespace types {
         LiveClientRealtimeInput,
         LiveClientToolResponse,
         LiveClientMessage,
+        StreamTranslationConfig,
         LiveConnectConfig,
         LiveConnectParameters,
         CreateChatParameters,
@@ -12680,30 +12728,6 @@ export declare interface VideoMetadata {
     startOffset?: string;
 }
 
-/**
- * Configuration for video output format.
- */
-declare interface VideoResponseFormat {
-    type: 'video';
-    /**
-     * The aspect ratio for the video output.
-     */
-    aspectRatio?: '16:9' | '9:16';
-    /**
-     * The delivery mode for the video output.
-     */
-    delivery?: 'inline' | 'uri';
-    /**
-     * The duration for the video output.
-     */
-    duration?: string;
-    /**
-     * The GCS URI to store the video output. Required for Vertex if delivery mode is
-     * URI.
-     */
-    gcsUri?: string;
-}
-
 /** Voice activity signal. */
 export declare interface VoiceActivity {
     /** The type of the voice activity signal. */