@google/genai 1.34.0 → 1.36.0

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
@@ -417,11 +417,17 @@ export declare interface AudioChunk {
417
417
  */
418
418
  declare interface AudioContent {
419
419
  type: 'audio';
420
+ /**
421
+ * The audio content.
422
+ */
420
423
  data?: string;
421
424
  /**
422
425
  * The mime type of the audio.
423
426
  */
424
427
  mime_type?: AudioMimeType;
428
+ /**
429
+ * The URI of the audio.
430
+ */
425
431
  uri?: string;
426
432
  }
427
433
 
@@ -581,7 +587,7 @@ declare interface BaseCreateAgentInteractionParams {
581
587
  /**
582
588
  * Body param: The inputs for the interaction.
583
589
  */
584
- input: string | Array<TextContent | ImageContent | AudioContent | DocumentContent | VideoContent | ThoughtContent | FunctionCallContent | FunctionResultContent | CodeExecutionCallContent | CodeExecutionResultContent | URLContextCallContent | URLContextResultContent | GoogleSearchCallContent | GoogleSearchResultContent | MCPServerToolCallContent | MCPServerToolResultContent | FileSearchResultContent> | Array<Turn> | TextContent | ImageContent | AudioContent | DocumentContent | VideoContent | ThoughtContent | FunctionCallContent | FunctionResultContent | CodeExecutionCallContent | CodeExecutionResultContent | URLContextCallContent | URLContextResultContent | GoogleSearchCallContent | GoogleSearchResultContent | MCPServerToolCallContent | MCPServerToolResultContent | FileSearchResultContent;
590
+ input: string | Array<Content_2> | Array<Turn> | TextContent | ImageContent | AudioContent | DocumentContent | VideoContent | ThoughtContent | FunctionCallContent | FunctionResultContent | CodeExecutionCallContent | CodeExecutionResultContent | URLContextCallContent | URLContextResultContent | GoogleSearchCallContent | GoogleSearchResultContent | MCPServerToolCallContent | MCPServerToolResultContent | FileSearchCallContent | FileSearchResultContent;
585
591
  /**
586
592
  * Path param: Which version of the API to use.
587
593
  */
@@ -633,7 +639,7 @@ declare interface BaseCreateModelInteractionParams {
633
639
  /**
634
640
  * Body param: The inputs for the interaction.
635
641
  */
636
- input: string | Array<TextContent | ImageContent | AudioContent | DocumentContent | VideoContent | ThoughtContent | FunctionCallContent | FunctionResultContent | CodeExecutionCallContent | CodeExecutionResultContent | URLContextCallContent | URLContextResultContent | GoogleSearchCallContent | GoogleSearchResultContent | MCPServerToolCallContent | MCPServerToolResultContent | FileSearchResultContent> | Array<Turn> | TextContent | ImageContent | AudioContent | DocumentContent | VideoContent | ThoughtContent | FunctionCallContent | FunctionResultContent | CodeExecutionCallContent | CodeExecutionResultContent | URLContextCallContent | URLContextResultContent | GoogleSearchCallContent | GoogleSearchResultContent | MCPServerToolCallContent | MCPServerToolResultContent | FileSearchResultContent;
642
+ input: string | Array<Content_2> | Array<Turn> | TextContent | ImageContent | AudioContent | DocumentContent | VideoContent | ThoughtContent | FunctionCallContent | FunctionResultContent | CodeExecutionCallContent | CodeExecutionResultContent | URLContextCallContent | URLContextResultContent | GoogleSearchCallContent | GoogleSearchResultContent | MCPServerToolCallContent | MCPServerToolResultContent | FileSearchCallContent | FileSearchResultContent;
637
643
  /**
638
644
  * Body param: The name of the `Model` used for generating the interaction.
639
645
  */
@@ -1045,7 +1051,7 @@ export declare interface BatchJobSource {
1045
1051
 
1046
1052
  export declare type BatchJobSourceUnion = BatchJobSource | InlinedRequest[] | string;
1047
1053
 
1048
- /** Defines the function behavior. Defaults to `BLOCKING`. */
1054
+ /** Specifies the function Behavior. Currently only supported by the BidiGenerateContent method. This enum is not supported in Vertex AI. */
1049
1055
  export declare enum Behavior {
1050
1056
  /**
1051
1057
  * This value is unused.
@@ -1701,6 +1707,12 @@ export declare class ComputeTokensResponse {
1701
1707
  tokensInfo?: TokensInfo[];
1702
1708
  }
1703
1709
 
1710
+ /** Local tokenizer compute tokens result. */
1711
+ export declare interface ComputeTokensResult {
1712
+ /** Lists of tokens info from the input. */
1713
+ tokensInfo?: TokensInfo[];
1714
+ }
1715
+
1704
1716
  declare type _ConditionalNodeReadableStream<R = any> = typeof globalThis extends {
1705
1717
  ReadableStream: any;
1706
1718
  } ? never : _NodeReadableStream<R>;
@@ -1717,11 +1729,15 @@ export declare interface Content {
1717
1729
  role?: string;
1718
1730
  }
1719
1731
 
1732
+ /**
1733
+ * The content of the response.
1734
+ */
1735
+ declare type Content_2 = TextContent | ImageContent | AudioContent | DocumentContent | VideoContent | ThoughtContent | FunctionCallContent | FunctionResultContent | CodeExecutionCallContent | CodeExecutionResultContent | URLContextCallContent | URLContextResultContent | GoogleSearchCallContent | GoogleSearchResultContent | MCPServerToolCallContent | MCPServerToolResultContent | FileSearchCallContent | FileSearchResultContent;
1736
+
1720
1737
  declare interface ContentDelta {
1721
- delta?: ContentDelta.TextDelta | ContentDelta.ImageDelta | ContentDelta.AudioDelta | ContentDelta.DocumentDelta | ContentDelta.VideoDelta | ContentDelta.ThoughtSummaryDelta | ContentDelta.ThoughtSignatureDelta | ContentDelta.FunctionCallDelta | ContentDelta.FunctionResultDelta | ContentDelta.CodeExecutionCallDelta | ContentDelta.CodeExecutionResultDelta | ContentDelta.URLContextCallDelta | ContentDelta.URLContextResultDelta | ContentDelta.GoogleSearchCallDelta | ContentDelta.GoogleSearchResultDelta | ContentDelta.MCPServerToolCallDelta | ContentDelta.MCPServerToolResultDelta | ContentDelta.FileSearchResultDelta;
1738
+ delta?: ContentDelta.TextDelta | ContentDelta.ImageDelta | ContentDelta.AudioDelta | ContentDelta.DocumentDelta | ContentDelta.VideoDelta | ContentDelta.ThoughtSummaryDelta | ContentDelta.ThoughtSignatureDelta | ContentDelta.FunctionCallDelta | ContentDelta.FunctionResultDelta | ContentDelta.CodeExecutionCallDelta | ContentDelta.CodeExecutionResultDelta | ContentDelta.URLContextCallDelta | ContentDelta.URLContextResultDelta | ContentDelta.GoogleSearchCallDelta | ContentDelta.GoogleSearchResultDelta | ContentDelta.MCPServerToolCallDelta | ContentDelta.MCPServerToolResultDelta | ContentDelta.FileSearchCallDelta | ContentDelta.FileSearchResultDelta;
1722
1739
  /**
1723
- * The event_id token to be used to resume the interaction stream, from
1724
- * this event.
1740
+ * The event_id token to be used to resume the interaction stream, from this event.
1725
1741
  */
1726
1742
  event_id?: string;
1727
1743
  event_type?: 'content.delta';
@@ -1917,6 +1933,13 @@ declare namespace ContentDelta {
1917
1933
  items?: Array<string | InteractionsAPI.ImageContent | unknown>;
1918
1934
  }
1919
1935
  }
1936
+ interface FileSearchCallDelta {
1937
+ type: 'file_search_call';
1938
+ /**
1939
+ * A unique ID for this specific tool call.
1940
+ */
1941
+ id?: string;
1942
+ }
1920
1943
  interface FileSearchResultDelta {
1921
1944
  type: 'file_search_result';
1922
1945
  result?: Array<FileSearchResultDelta.Result>;
@@ -1986,10 +2009,9 @@ declare interface ContentStart {
1986
2009
  /**
1987
2010
  * The content of the response.
1988
2011
  */
1989
- content?: TextContent | ImageContent | AudioContent | DocumentContent | VideoContent | ThoughtContent | FunctionCallContent | FunctionResultContent | CodeExecutionCallContent | CodeExecutionResultContent | URLContextCallContent | URLContextResultContent | GoogleSearchCallContent | GoogleSearchResultContent | MCPServerToolCallContent | MCPServerToolResultContent | FileSearchResultContent;
2012
+ content?: Content_2;
1990
2013
  /**
1991
- * The event_id token to be used to resume the interaction stream, from
1992
- * this event.
2014
+ * The event_id token to be used to resume the interaction stream, from this event.
1993
2015
  */
1994
2016
  event_id?: string;
1995
2017
  event_type?: 'content.start';
@@ -1998,8 +2020,7 @@ declare interface ContentStart {
1998
2020
 
1999
2021
  declare interface ContentStop {
2000
2022
  /**
2001
- * The event_id token to be used to resume the interaction stream, from
2002
- * this event.
2023
+ * The event_id token to be used to resume the interaction stream, from this event.
2003
2024
  */
2004
2025
  event_id?: string;
2005
2026
  event_type?: 'content.stop';
@@ -2101,6 +2122,12 @@ export declare class CountTokensResponse {
2101
2122
  cachedContentTokenCount?: number;
2102
2123
  }
2103
2124
 
2125
+ /** Local tokenizer count tokens result. */
2126
+ export declare interface CountTokensResult {
2127
+ /** The total number of tokens. */
2128
+ totalTokens?: number;
2129
+ }
2130
+
2104
2131
  declare interface CreateAgentInteractionParamsNonStreaming extends BaseCreateAgentInteractionParams {
2105
2132
  /**
2106
2133
  * Body param: Input only. Whether the interaction will be streamed.
@@ -2752,11 +2779,17 @@ export { Document_2 as Document }
2752
2779
  */
2753
2780
  declare interface DocumentContent {
2754
2781
  type: 'document';
2782
+ /**
2783
+ * The document content.
2784
+ */
2755
2785
  data?: string;
2756
2786
  /**
2757
2787
  * The mime type of the document.
2758
2788
  */
2759
2789
  mime_type?: DocumentMimeType;
2790
+ /**
2791
+ * The URI of the document.
2792
+ */
2760
2793
  uri?: string;
2761
2794
  }
2762
2795
 
@@ -2856,13 +2889,13 @@ declare interface DynamicAgentConfig {
2856
2889
 
2857
2890
  /** Describes the options to customize dynamic retrieval. */
2858
2891
  export declare interface DynamicRetrievalConfig {
2859
- /** The mode of the predictor to be used in dynamic retrieval. */
2860
- mode?: DynamicRetrievalConfigMode;
2861
2892
  /** Optional. The threshold to be used in dynamic retrieval. If not set, a system default value is used. */
2862
2893
  dynamicThreshold?: number;
2894
+ /** The mode of the predictor to be used in dynamic retrieval. */
2895
+ mode?: DynamicRetrievalConfigMode;
2863
2896
  }
2864
2897
 
2865
- /** Config for the dynamic retrieval config mode. */
2898
+ /** The mode of the predictor to be used in dynamic retrieval. */
2866
2899
  export declare enum DynamicRetrievalConfigMode {
2867
2900
  /**
2868
2901
  * Always trigger retrieval.
@@ -3113,8 +3146,7 @@ declare interface ErrorEvent_2 {
3113
3146
  */
3114
3147
  error?: ErrorEvent_2.Error;
3115
3148
  /**
3116
- * The event_id token to be used to resume the interaction stream, from
3117
- * this event.
3149
+ * The event_id token to be used to resume the interaction stream, from this event.
3118
3150
  */
3119
3151
  event_id?: string;
3120
3152
  event_type?: 'error';
@@ -3398,6 +3430,17 @@ export declare interface FileSearch {
3398
3430
  metadataFilter?: string;
3399
3431
  }
3400
3432
 
3433
+ /**
3434
+ * File Search content.
3435
+ */
3436
+ declare interface FileSearchCallContent {
3437
+ type: 'file_search_call';
3438
+ /**
3439
+ * A unique ID for this specific tool call.
3440
+ */
3441
+ id?: string;
3442
+ }
3443
+
3401
3444
  /**
3402
3445
  * File Search result content.
3403
3446
  */
@@ -3543,7 +3586,8 @@ declare class FileSearchStores extends BaseModule {
3543
3586
  export declare enum FileSource {
3544
3587
  SOURCE_UNSPECIFIED = "SOURCE_UNSPECIFIED",
3545
3588
  UPLOADED = "UPLOADED",
3546
- GENERATED = "GENERATED"
3589
+ GENERATED = "GENERATED",
3590
+ REGISTERED = "REGISTERED"
3547
3591
  }
3548
3592
 
3549
3593
  /**
@@ -3721,18 +3765,18 @@ declare interface FunctionCallContent {
3721
3765
 
3722
3766
  /** Function calling config. */
3723
3767
  export declare interface FunctionCallingConfig {
3724
- /** Optional. Function calling mode. */
3725
- mode?: FunctionCallingConfigMode;
3726
3768
  /** Optional. Function names to call. Only set when the Mode is ANY. Function names should match [FunctionDeclaration.name]. With mode set to ANY, model will predict a function call from the set of function names provided. */
3727
3769
  allowedFunctionNames?: string[];
3770
+ /** Optional. Function calling mode. */
3771
+ mode?: FunctionCallingConfigMode;
3728
3772
  /** Optional. When set to true, arguments of a single function call will be streamed out in multiple parts/contents/responses. Partial parameter results will be returned in the [FunctionCall.partial_args] field. This field is not supported in Gemini API. */
3729
3773
  streamFunctionCallArguments?: boolean;
3730
3774
  }
3731
3775
 
3732
- /** Config for the function calling config mode. */
3776
+ /** Function calling mode. */
3733
3777
  export declare enum FunctionCallingConfigMode {
3734
3778
  /**
3735
- * The function calling config mode is unspecified. Should not be used.
3779
+ * Unspecified function calling mode. This value should not be used.
3736
3780
  */
3737
3781
  MODE_UNSPECIFIED = "MODE_UNSPECIFIED",
3738
3782
  /**
@@ -3748,18 +3792,13 @@ export declare enum FunctionCallingConfigMode {
3748
3792
  */
3749
3793
  NONE = "NONE",
3750
3794
  /**
3751
- * Model decides to predict either a function call or a natural language response, but will validate function calls with constrained decoding. If "allowed_function_names" are set, the predicted function call will be limited to any one of "allowed_function_names", else the predicted function call will be any one of the provided "function_declarations".
3795
+ * Model is constrained to predict either function calls or natural language response. If "allowed_function_names" are set, the predicted function calls will be limited to any one of "allowed_function_names", else the predicted function calls will be any one of the provided "function_declarations".
3752
3796
  */
3753
3797
  VALIDATED = "VALIDATED"
3754
3798
  }
3755
3799
 
3756
- /** Defines a function that the model can generate JSON inputs for.
3757
-
3758
- The inputs are based on `OpenAPI 3.0 specifications
3759
- <https://spec.openapis.org/oas/v3.0.3>`_. */
3800
+ /** Structured representation of a function declaration as defined by the [OpenAPI 3.0 specification](https://spec.openapis.org/oas/v3.0.3). Included in this declaration are the function name, description, parameters and response type. This FunctionDeclaration is a representation of a block of code that can be used as a `Tool` by the model and executed by the client. */
3760
3801
  export declare interface FunctionDeclaration {
3761
- /** Defines the function behavior. */
3762
- behavior?: Behavior;
3763
3802
  /** Optional. Description and purpose of the function. Model uses it to decide how and whether to call the function. */
3764
3803
  description?: string;
3765
3804
  /** Required. The name of the function to call. Must start with a letter or an underscore. Must be a-z, A-Z, 0-9, or contain underscores, dots and dashes, with a maximum length of 64. */
@@ -3772,6 +3811,8 @@ export declare interface FunctionDeclaration {
3772
3811
  response?: Schema;
3773
3812
  /** Optional. Describes the output from this function in JSON Schema format. The value specified by the schema is the response value of the function. This field is mutually exclusive with `response`. */
3774
3813
  responseJsonSchema?: unknown;
3814
+ /** Optional. Specifies the function Behavior. Currently only supported by the BidiGenerateContent method. This field is not supported in Vertex AI. */
3815
+ behavior?: Behavior;
3775
3816
  }
3776
3817
 
3777
3818
  /** A function response. */
@@ -4561,6 +4602,10 @@ export declare interface GenerationConfig {
4561
4602
  * Configuration parameters for model interactions.
4562
4603
  */
4563
4604
  declare interface GenerationConfig_2 {
4605
+ /**
4606
+ * Configuration for image interaction.
4607
+ */
4608
+ image_config?: ImageConfig_2;
4564
4609
  /**
4565
4610
  * The maximum number of tokens to include in the response.
4566
4611
  */
@@ -5433,6 +5478,9 @@ export declare interface ImageConfig {
5433
5478
  values are `1K`, `2K`, `4K`. If not specified, the model will use default
5434
5479
  value `1K`. */
5435
5480
  imageSize?: string;
5481
+ /** Controls the generation of people. Supported values are:
5482
+ ALLOW_ALL, ALLOW_ADULT, ALLOW_NONE. */
5483
+ personGeneration?: string;
5436
5484
  /** MIME type of the generated image. This field is not
5437
5485
  supported in Gemini API. */
5438
5486
  outputMimeType?: string;
@@ -5441,11 +5489,22 @@ export declare interface ImageConfig {
5441
5489
  outputCompressionQuality?: number;
5442
5490
  }
5443
5491
 
5492
+ /**
5493
+ * The configuration for image interaction.
5494
+ */
5495
+ declare interface ImageConfig_2 {
5496
+ aspect_ratio?: '1:1' | '2:3' | '3:2' | '3:4' | '4:3' | '9:16' | '16:9' | '21:9';
5497
+ image_size?: '1K' | '2K' | '4K';
5498
+ }
5499
+
5444
5500
  /**
5445
5501
  * An image content block.
5446
5502
  */
5447
5503
  declare interface ImageContent {
5448
5504
  type: 'image';
5505
+ /**
5506
+ * The image content.
5507
+ */
5449
5508
  data?: string;
5450
5509
  /**
5451
5510
  * The mime type of the image.
@@ -5455,6 +5514,9 @@ declare interface ImageContent {
5455
5514
  * The resolution of the media.
5456
5515
  */
5457
5516
  resolution?: 'low' | 'medium' | 'high' | 'ultra_high';
5517
+ /**
5518
+ * The URI of the image.
5519
+ */
5458
5520
  uri?: string;
5459
5521
  }
5460
5522
 
@@ -5617,14 +5679,10 @@ declare interface Interaction {
5617
5679
  * The name of the `Model` used for generating the interaction.
5618
5680
  */
5619
5681
  model?: Model_2;
5620
- /**
5621
- * Output only. The object type of the interaction. Always set to `interaction`.
5622
- */
5623
- object?: 'interaction';
5624
5682
  /**
5625
5683
  * Output only. Responses from the model.
5626
5684
  */
5627
- outputs?: Array<TextContent | ImageContent | AudioContent | DocumentContent | VideoContent | ThoughtContent | FunctionCallContent | FunctionResultContent | CodeExecutionCallContent | CodeExecutionResultContent | URLContextCallContent | URLContextResultContent | GoogleSearchCallContent | GoogleSearchResultContent | MCPServerToolCallContent | MCPServerToolResultContent | FileSearchResultContent>;
5685
+ outputs?: Array<Content_2>;
5628
5686
  /**
5629
5687
  * The ID of the previous interaction, if any.
5630
5688
  */
@@ -5664,8 +5722,7 @@ declare type InteractionDeleteResponse = unknown;
5664
5722
 
5665
5723
  declare interface InteractionEvent {
5666
5724
  /**
5667
- * The event_id token to be used to resume the interaction stream, from
5668
- * this event.
5725
+ * The event_id token to be used to resume the interaction stream, from this event.
5669
5726
  */
5670
5727
  event_id?: string;
5671
5728
  event_type?: 'interaction.start' | 'interaction.complete';
@@ -5715,7 +5772,7 @@ export declare class Interactions extends BaseInteractions {
5715
5772
  }
5716
5773
 
5717
5774
  export declare namespace Interactions {
5718
- export { type AllowedTools as AllowedTools, type Annotation as Annotation, type AudioContent as AudioContent, type AudioMimeType as AudioMimeType, type CodeExecutionCallArguments as CodeExecutionCallArguments, type CodeExecutionCallContent as CodeExecutionCallContent, type CodeExecutionResultContent as CodeExecutionResultContent, type ContentDelta as ContentDelta, type ContentStart as ContentStart, type ContentStop as ContentStop, type DeepResearchAgentConfig as DeepResearchAgentConfig, type DocumentContent as DocumentContent, type DocumentMimeType as DocumentMimeType, type DynamicAgentConfig as DynamicAgentConfig, type ErrorEvent_2 as ErrorEvent, type FileSearchResultContent as FileSearchResultContent, type Function_2 as Function, type FunctionCallContent as FunctionCallContent, type FunctionResultContent as FunctionResultContent, type GenerationConfig_2 as GenerationConfig, type GoogleSearchCallArguments as GoogleSearchCallArguments, type GoogleSearchCallContent as GoogleSearchCallContent, type GoogleSearchResult as GoogleSearchResult, type GoogleSearchResultContent as GoogleSearchResultContent, type ImageContent as ImageContent, type ImageMimeType as ImageMimeType, type Interaction as Interaction, type InteractionEvent as InteractionEvent, type InteractionSSEEvent as InteractionSSEEvent, type InteractionStatusUpdate as InteractionStatusUpdate, type MCPServerToolCallContent as MCPServerToolCallContent, type MCPServerToolResultContent as MCPServerToolResultContent, type Model_2 as Model, type SpeechConfig_2 as SpeechConfig, type TextContent as TextContent, type ThinkingLevel_2 as ThinkingLevel, type ThoughtContent as ThoughtContent, type Tool_2 as Tool, type ToolChoice as ToolChoice, type ToolChoiceConfig as ToolChoiceConfig, type ToolChoiceType as ToolChoiceType, type Turn as Turn, type URLContextCallArguments as URLContextCallArguments, type URLContextCallContent as URLContextCallContent, type URLContextResult as URLContextResult, type URLContextResultContent as URLContextResultContent, type Usage as Usage, type VideoContent as VideoContent, type VideoMimeType as VideoMimeType, 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, };
5775
+ export { type AllowedTools as AllowedTools, type Annotation as Annotation, type AudioContent as AudioContent, type AudioMimeType as AudioMimeType, type CodeExecutionCallArguments as CodeExecutionCallArguments, type CodeExecutionCallContent as CodeExecutionCallContent, type CodeExecutionResultContent as CodeExecutionResultContent, type Content_2 as Content, type ContentDelta as ContentDelta, type ContentStart as ContentStart, type ContentStop as ContentStop, type DeepResearchAgentConfig as DeepResearchAgentConfig, type DocumentContent as DocumentContent, type DocumentMimeType as DocumentMimeType, type DynamicAgentConfig as DynamicAgentConfig, type ErrorEvent_2 as ErrorEvent, type FileSearchCallContent as FileSearchCallContent, type FileSearchResultContent as FileSearchResultContent, type Function_2 as Function, type FunctionCallContent as FunctionCallContent, type FunctionResultContent as FunctionResultContent, type GenerationConfig_2 as GenerationConfig, type GoogleSearchCallArguments as GoogleSearchCallArguments, type GoogleSearchCallContent as GoogleSearchCallContent, type GoogleSearchResult as GoogleSearchResult, type GoogleSearchResultContent as GoogleSearchResultContent, type ImageConfig_2 as ImageConfig, type ImageContent as ImageContent, type ImageMimeType as ImageMimeType, type Interaction as Interaction, type InteractionEvent as InteractionEvent, type InteractionSSEEvent as InteractionSSEEvent, type InteractionStatusUpdate as InteractionStatusUpdate, type MCPServerToolCallContent as MCPServerToolCallContent, type MCPServerToolResultContent as MCPServerToolResultContent, type Model_2 as Model, type SpeechConfig_2 as SpeechConfig, type TextContent as TextContent, type ThinkingLevel_2 as ThinkingLevel, type ThoughtContent as ThoughtContent, type Tool_2 as Tool, type ToolChoice as ToolChoice, type ToolChoiceConfig as ToolChoiceConfig, type ToolChoiceType as ToolChoiceType, type Turn as Turn, type URLContextCallArguments as URLContextCallArguments, type URLContextCallContent as URLContextCallContent, type URLContextResult as URLContextResult, type URLContextResultContent as URLContextResultContent, type Usage as Usage, type VideoContent as VideoContent, type VideoMimeType as VideoMimeType, 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, };
5719
5776
  }
5720
5777
 
5721
5778
  declare namespace InteractionsAPI {
@@ -5729,6 +5786,7 @@ declare namespace InteractionsAPI {
5729
5786
  CodeExecutionCallArguments,
5730
5787
  CodeExecutionCallContent,
5731
5788
  CodeExecutionResultContent,
5789
+ Content_2 as Content,
5732
5790
  ContentDelta,
5733
5791
  ContentStart,
5734
5792
  ContentStop,
@@ -5737,6 +5795,7 @@ declare namespace InteractionsAPI {
5737
5795
  DocumentMimeType,
5738
5796
  DynamicAgentConfig,
5739
5797
  ErrorEvent_2 as ErrorEvent,
5798
+ FileSearchCallContent,
5740
5799
  FileSearchResultContent,
5741
5800
  Function_2 as Function,
5742
5801
  FunctionCallContent,
@@ -5746,6 +5805,7 @@ declare namespace InteractionsAPI {
5746
5805
  GoogleSearchCallContent,
5747
5806
  GoogleSearchResult,
5748
5807
  GoogleSearchResultContent,
5808
+ ImageConfig_2 as ImageConfig,
5749
5809
  ImageContent,
5750
5810
  ImageMimeType,
5751
5811
  Interaction,
@@ -5792,8 +5852,7 @@ declare type InteractionSSEEvent = InteractionEvent | InteractionStatusUpdate |
5792
5852
 
5793
5853
  declare interface InteractionStatusUpdate {
5794
5854
  /**
5795
- * The event_id token to be used to resume the interaction stream, from
5796
- * this event.
5855
+ * The event_id token to be used to resume the interaction stream, from this event.
5797
5856
  */
5798
5857
  event_id?: string;
5799
5858
  event_type?: 'interaction.status_update';
@@ -6820,8 +6879,10 @@ export declare class LiveServerMessage {
6820
6879
  goAway?: LiveServerGoAway;
6821
6880
  /** Update of the session resumption state. */
6822
6881
  sessionResumptionUpdate?: LiveServerSessionResumptionUpdate;
6823
- /** Voice activity detection signal. */
6882
+ /** Voice activity detection signal. Allowlisted only. */
6824
6883
  voiceActivityDetectionSignal?: VoiceActivityDetectionSignal;
6884
+ /** Voice activity signal. */
6885
+ voiceActivity?: VoiceActivity;
6825
6886
  /**
6826
6887
  * Returns the concatenation of all text parts from the server content if present.
6827
6888
  *
@@ -7108,18 +7169,6 @@ export declare interface ModalityTokenCount {
7108
7169
  tokenCount?: number;
7109
7170
  }
7110
7171
 
7111
- /** The mode of the predictor to be used in dynamic retrieval. */
7112
- export declare enum Mode {
7113
- /**
7114
- * Always trigger retrieval.
7115
- */
7116
- MODE_UNSPECIFIED = "MODE_UNSPECIFIED",
7117
- /**
7118
- * Run retrieval only when system decides it is necessary.
7119
- */
7120
- MODE_DYNAMIC = "MODE_DYNAMIC"
7121
- }
7122
-
7123
7172
  /** A trained machine learning model. */
7124
7173
  export declare interface Model {
7125
7174
  /** Resource name of the model. */
@@ -7178,7 +7227,7 @@ export declare interface Model {
7178
7227
  /**
7179
7228
  * The model that will complete your prompt.\n\nSee [models](https://ai.google.dev/gemini-api/docs/models) for additional details.
7180
7229
  */
7181
- declare type Model_2 = 'gemini-2.5-pro' | 'gemini-2.5-flash' | 'gemini-2.5-flash-preview-09-2025' | 'gemini-2.5-flash-lite' | 'gemini-2.5-flash-lite-preview-09-2025' | 'gemini-2.5-flash-preview-native-audio-dialog' | 'gemini-2.5-flash-image-preview' | 'gemini-2.5-pro-preview-tts' | 'gemini-3-pro-preview' | (string & {});
7230
+ declare type Model_2 = 'gemini-2.5-pro' | 'gemini-2.5-flash' | 'gemini-2.5-flash-preview-09-2025' | 'gemini-2.5-flash-lite' | 'gemini-2.5-flash-lite-preview-09-2025' | 'gemini-2.5-flash-preview-native-audio-dialog' | 'gemini-2.5-flash-image-preview' | 'gemini-2.5-pro-preview-tts' | 'gemini-3-pro-preview' | 'gemini-3-flash-preview' | (string & {});
7182
7231
 
7183
7232
  export declare class Models extends BaseModule {
7184
7233
  private readonly apiClient;
@@ -7290,7 +7339,7 @@ export declare class Models extends BaseModule {
7290
7339
  * @example
7291
7340
  * ```ts
7292
7341
  * const response = await client.models.generateImages({
7293
- * model: 'imagen-3.0-generate-002',
7342
+ * model: 'imagen-4.0-generate-001',
7294
7343
  * prompt: 'Robot holding a red skateboard',
7295
7344
  * config: {
7296
7345
  * numberOfImages: 1,
@@ -7333,7 +7382,7 @@ export declare class Models extends BaseModule {
7333
7382
  * @example
7334
7383
  * ```ts
7335
7384
  * const response = await client.models.upscaleImage({
7336
- * model: 'imagen-3.0-generate-002',
7385
+ * model: 'imagen-4.0-upscale-preview',
7337
7386
  * image: image,
7338
7387
  * upscaleFactor: 'x2',
7339
7388
  * config: {
@@ -7432,7 +7481,7 @@ export declare class Models extends BaseModule {
7432
7481
  * console.log(response1?.generatedImages?.[0]?.image?.imageBytes);
7433
7482
  *
7434
7483
  * const response2 = await ai.models.recontextImage({
7435
- * model: 'virtual-try-on-preview-08-04',
7484
+ * model: 'virtual-try-on-001',
7436
7485
  * source: {
7437
7486
  * personImage: personImage,
7438
7487
  * productImages: [productImage],
@@ -9287,12 +9336,8 @@ export declare interface TokensInfo {
9287
9336
 
9288
9337
  /** Tool details of a tool that the model may use to generate a response. */
9289
9338
  export declare interface Tool {
9290
- /** List of function declarations that the tool supports. */
9291
- functionDeclarations?: FunctionDeclaration[];
9292
9339
  /** Optional. Retrieval tool type. System will always execute the provided retrieval tool(s) to get external knowledge to answer the prompt. Retrieval results are presented to the model for generation. This field is not supported in Gemini API. */
9293
9340
  retrieval?: Retrieval;
9294
- /** Optional. Specialized retrieval tool that is powered by Google Search. */
9295
- googleSearchRetrieval?: GoogleSearchRetrieval;
9296
9341
  /** Optional. Tool to support the model interacting directly with the
9297
9342
  computer. If enabled, it automatically populates computer-use specific
9298
9343
  Function Declarations. */
@@ -9303,10 +9348,14 @@ export declare interface Tool {
9303
9348
  codeExecution?: ToolCodeExecution;
9304
9349
  /** Optional. Tool to support searching public web data, powered by Vertex AI Search and Sec4 compliance. This field is not supported in Gemini API. */
9305
9350
  enterpriseWebSearch?: EnterpriseWebSearch;
9351
+ /** Optional. Function tool type. One or more function declarations to be passed to the model along with the current user query. Model may decide to call a subset of these functions by populating FunctionCall in the response. User should provide a FunctionResponse for each function call in the next turn. Based on the function responses, Model will generate the final response back to the user. Maximum 512 function declarations can be provided. */
9352
+ functionDeclarations?: FunctionDeclaration[];
9306
9353
  /** Optional. GoogleMaps tool type. Tool to support Google Maps in Model. */
9307
9354
  googleMaps?: GoogleMaps;
9308
9355
  /** Optional. GoogleSearch tool type. Tool to support Google Search in Model. Powered by Google. */
9309
9356
  googleSearch?: GoogleSearch;
9357
+ /** Optional. Specialized retrieval tool that is powered by Google Search. */
9358
+ googleSearchRetrieval?: GoogleSearchRetrieval;
9310
9359
  /** Optional. Tool to support URL context retrieval. */
9311
9360
  urlContext?: UrlContext;
9312
9361
  }
@@ -9416,10 +9465,10 @@ export declare interface ToolCodeExecution {
9416
9465
 
9417
9466
  This config is shared for all tools provided in the request. */
9418
9467
  export declare interface ToolConfig {
9419
- /** Optional. Function calling config. */
9420
- functionCallingConfig?: FunctionCallingConfig;
9421
9468
  /** Optional. Retrieval config. */
9422
9469
  retrievalConfig?: RetrievalConfig;
9470
+ /** Optional. Function calling config. */
9471
+ functionCallingConfig?: FunctionCallingConfig;
9423
9472
  }
9424
9473
 
9425
9474
  export declare type ToolListUnion = ToolUnion[];
@@ -9709,7 +9758,7 @@ declare interface Turn {
9709
9758
  /**
9710
9759
  * The content of the turn.
9711
9760
  */
9712
- content?: string | Array<TextContent | ImageContent | AudioContent | DocumentContent | VideoContent | ThoughtContent | FunctionCallContent | FunctionResultContent | CodeExecutionCallContent | CodeExecutionResultContent | URLContextCallContent | URLContextResultContent | GoogleSearchCallContent | GoogleSearchResultContent | MCPServerToolCallContent | MCPServerToolResultContent | FileSearchResultContent>;
9761
+ content?: string | Array<Content_2>;
9713
9762
  /**
9714
9763
  * The originator of this turn. Must be user for input or model for
9715
9764
  * model output.
@@ -9806,11 +9855,13 @@ declare namespace types {
9806
9855
  Language,
9807
9856
  FunctionResponseScheduling,
9808
9857
  Type,
9809
- Mode,
9810
9858
  ApiSpec,
9811
9859
  AuthType,
9812
9860
  HttpElementLocation,
9813
9861
  PhishBlockThreshold,
9862
+ Behavior,
9863
+ DynamicRetrievalConfigMode,
9864
+ FunctionCallingConfigMode,
9814
9865
  ThinkingLevel,
9815
9866
  HarmCategory,
9816
9867
  HarmBlockMethod,
@@ -9829,10 +9880,7 @@ declare namespace types {
9829
9880
  TuningTask,
9830
9881
  PartMediaResolutionLevel,
9831
9882
  FeatureSelectionPreference,
9832
- Behavior,
9833
- DynamicRetrievalConfigMode,
9834
9883
  Environment,
9835
- FunctionCallingConfigMode,
9836
9884
  SafetyFilterLevel,
9837
9885
  PersonGeneration,
9838
9886
  ImagePromptLanguage,
@@ -9851,6 +9899,7 @@ declare namespace types {
9851
9899
  TurnCompleteReason,
9852
9900
  MediaModality,
9853
9901
  VadSignalType,
9902
+ VoiceActivityType,
9854
9903
  StartSensitivity,
9855
9904
  EndSensitivity,
9856
9905
  ActivityHandling,
@@ -9875,9 +9924,6 @@ declare namespace types {
9875
9924
  HttpOptions,
9876
9925
  Schema,
9877
9926
  ModelSelectionConfig,
9878
- FunctionDeclaration,
9879
- DynamicRetrievalConfig,
9880
- GoogleSearchRetrieval,
9881
9927
  ComputerUse,
9882
9928
  FileSearch,
9883
9929
  ApiAuthApiKeyConfig,
@@ -9904,14 +9950,17 @@ declare namespace types {
9904
9950
  Retrieval,
9905
9951
  ToolCodeExecution,
9906
9952
  EnterpriseWebSearch,
9953
+ FunctionDeclaration,
9907
9954
  GoogleMaps,
9908
9955
  Interval,
9909
9956
  GoogleSearch,
9957
+ DynamicRetrievalConfig,
9958
+ GoogleSearchRetrieval,
9910
9959
  UrlContext,
9911
9960
  Tool,
9912
- FunctionCallingConfig,
9913
9961
  LatLng,
9914
9962
  RetrievalConfig,
9963
+ FunctionCallingConfig,
9915
9964
  ToolConfig,
9916
9965
  ReplicatedVoiceConfig,
9917
9966
  PrebuiltVoiceConfig,
@@ -10173,6 +10222,7 @@ declare namespace types {
10173
10222
  LiveServerGoAway,
10174
10223
  LiveServerSessionResumptionUpdate,
10175
10224
  VoiceActivityDetectionSignal,
10225
+ VoiceActivity,
10176
10226
  LiveServerMessage,
10177
10227
  OperationFromAPIResponseParameters,
10178
10228
  GenerationConfigThinkingConfig,
@@ -10220,6 +10270,8 @@ declare namespace types {
10220
10270
  CreateAuthTokenConfig,
10221
10271
  CreateAuthTokenParameters,
10222
10272
  OperationGetParameters,
10273
+ CountTokensResult,
10274
+ ComputeTokensResult,
10223
10275
  CreateTuningJobParameters,
10224
10276
  UploadToFileSearchStoreResponse,
10225
10277
  UploadToFileSearchStoreOperation,
@@ -10654,7 +10706,7 @@ declare interface Usage {
10654
10706
  /**
10655
10707
  * Number of tokens of thoughts for thinking models.
10656
10708
  */
10657
- total_reasoning_tokens?: number;
10709
+ total_thought_tokens?: number;
10658
10710
  /**
10659
10711
  * Total token count for the interaction request (prompt + responses + other
10660
10712
  * internal tokens).
@@ -10860,6 +10912,9 @@ export declare enum VideoCompressionQuality {
10860
10912
  */
10861
10913
  declare interface VideoContent {
10862
10914
  type: 'video';
10915
+ /**
10916
+ * The video content.
10917
+ */
10863
10918
  data?: string;
10864
10919
  /**
10865
10920
  * The mime type of the video.
@@ -10869,6 +10924,9 @@ declare interface VideoContent {
10869
10924
  * The resolution of the media.
10870
10925
  */
10871
10926
  resolution?: 'low' | 'medium' | 'high' | 'ultra_high';
10927
+ /**
10928
+ * The URI of the video.
10929
+ */
10872
10930
  uri?: string;
10873
10931
  }
10874
10932
 
@@ -10948,11 +11006,33 @@ export declare interface VideoMetadata {
10948
11006
  */
10949
11007
  declare type VideoMimeType = 'video/mp4' | 'video/mpeg' | 'video/mov' | 'video/avi' | 'video/x-flv' | 'video/mpg' | 'video/webm' | 'video/wmv' | 'video/3gpp' | (string & {});
10950
11008
 
11009
+ /** Voice activity signal. */
11010
+ export declare interface VoiceActivity {
11011
+ /** The type of the voice activity signal. */
11012
+ voiceActivityType?: VoiceActivityType;
11013
+ }
11014
+
10951
11015
  export declare interface VoiceActivityDetectionSignal {
10952
11016
  /** The type of the VAD signal. */
10953
11017
  vadSignalType?: VadSignalType;
10954
11018
  }
10955
11019
 
11020
+ /** The type of the voice activity signal. */
11021
+ export declare enum VoiceActivityType {
11022
+ /**
11023
+ * The default is VOICE_ACTIVITY_TYPE_UNSPECIFIED.
11024
+ */
11025
+ TYPE_UNSPECIFIED = "TYPE_UNSPECIFIED",
11026
+ /**
11027
+ * Start of sentence signal.
11028
+ */
11029
+ ACTIVITY_START = "ACTIVITY_START",
11030
+ /**
11031
+ * End of sentence signal.
11032
+ */
11033
+ ACTIVITY_END = "ACTIVITY_END"
11034
+ }
11035
+
10956
11036
  export declare interface VoiceConfig {
10957
11037
  /** If true, the model will use a replicated voice for the response. */
10958
11038
  replicatedVoiceConfig?: ReplicatedVoiceConfig;