@google/genai 1.20.0 → 1.22.0

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.
@@ -1069,6 +1069,18 @@ export declare interface CodeExecutionResult {
1069
1069
  output?: string;
1070
1070
  }
1071
1071
 
1072
+ /** Tool to support computer use. */
1073
+ export declare interface ComputerUse {
1074
+ /** Required. The environment being operated. */
1075
+ environment?: Environment;
1076
+ /** By default, predefined functions are included in the final model call.
1077
+ Some of them can be explicitly excluded from being automatically included.
1078
+ This can serve two purposes:
1079
+ 1. Using a more restricted / different action space.
1080
+ 2. Improving the definitions / instructions of predefined functions. */
1081
+ excludedPredefinedFunctions?: string[];
1082
+ }
1083
+
1072
1084
  /** Optional parameters for computing tokens. */
1073
1085
  export declare interface ComputeTokensConfig {
1074
1086
  /** Used to override HTTP request options. */
@@ -1137,6 +1149,23 @@ export declare interface ContentEmbeddingStatistics {
1137
1149
 
1138
1150
  export declare type ContentListUnion = Content | Content[] | PartUnion | PartUnion[];
1139
1151
 
1152
+ /** A content reference image.
1153
+
1154
+ A content reference image represents a subject to reference (ex. person,
1155
+ product, animal) provided by the user. It can optionally be provided in
1156
+ addition to a style reference image (ex. background, style reference).
1157
+ */
1158
+ export declare class ContentReferenceImage {
1159
+ /** The reference image for the editing operation. */
1160
+ referenceImage?: Image_2;
1161
+ /** The id of the reference image. */
1162
+ referenceId?: number;
1163
+ /** The type of the reference image. Only set by the SDK. */
1164
+ referenceType?: string;
1165
+ /** Internal method to convert to ReferenceImageAPIInternal. */
1166
+ toReferenceImageAPI(): ReferenceImageAPIInternal;
1167
+ }
1168
+
1140
1169
  export declare type ContentUnion = Content | PartUnion[] | PartUnion;
1141
1170
 
1142
1171
  /** Enables context window compression -- mechanism managing model context window so it does not exceed given length. */
@@ -2327,7 +2356,11 @@ export declare enum FinishReason {
2327
2356
  /**
2328
2357
  * The tool call generated by the model is invalid.
2329
2358
  */
2330
- UNEXPECTED_TOOL_CALL = "UNEXPECTED_TOOL_CALL"
2359
+ UNEXPECTED_TOOL_CALL = "UNEXPECTED_TOOL_CALL",
2360
+ /**
2361
+ * Image generation stopped because the generated images have prohibited content.
2362
+ */
2363
+ IMAGE_PROHIBITED_CONTENT = "IMAGE_PROHIBITED_CONTENT"
2331
2364
  }
2332
2365
 
2333
2366
  /** A function call. */
@@ -2401,6 +2434,9 @@ export declare class FunctionResponse {
2401
2434
  willContinue?: boolean;
2402
2435
  /** Specifies how the response should be scheduled in the conversation. Only applicable to NON_BLOCKING function calls, is ignored otherwise. Defaults to WHEN_IDLE. */
2403
2436
  scheduling?: FunctionResponseScheduling;
2437
+ /** List of parts that constitute a function response. Each part may
2438
+ have a different IANA MIME type. */
2439
+ parts?: FunctionResponsePart[];
2404
2440
  /** Optional. The id of the function call this response is for. Populated by the client to match the corresponding function call `id`. */
2405
2441
  id?: string;
2406
2442
  /** Required. The name of the function to call. Matches [FunctionDeclaration.name] and [FunctionCall.name]. */
@@ -2409,6 +2445,44 @@ export declare class FunctionResponse {
2409
2445
  response?: Record<string, unknown>;
2410
2446
  }
2411
2447
 
2448
+ /** Raw media bytes for function response.
2449
+
2450
+ Text should not be sent as raw bytes, use the FunctionResponse.response
2451
+ field.
2452
+ */
2453
+ export declare class FunctionResponseBlob {
2454
+ /** Required. The IANA standard MIME type of the source data. */
2455
+ mimeType?: string;
2456
+ /** Required. Inline media bytes.
2457
+ * @remarks Encoded as base64 string. */
2458
+ data?: string;
2459
+ }
2460
+
2461
+ /** URI based data for function response. */
2462
+ export declare class FunctionResponseFileData {
2463
+ /** Required. URI. */
2464
+ fileUri?: string;
2465
+ /** Required. The IANA standard MIME type of the source data. */
2466
+ mimeType?: string;
2467
+ }
2468
+
2469
+ /** A datatype containing media that is part of a `FunctionResponse` message.
2470
+
2471
+ A `FunctionResponsePart` consists of data which has an associated datatype. A
2472
+ `FunctionResponsePart` can only contain one of the accepted types in
2473
+ `FunctionResponsePart.data`.
2474
+
2475
+ A `FunctionResponsePart` must have a fixed IANA MIME type identifying the
2476
+ type and subtype of the media if the `inline_data` field is filled with raw
2477
+ bytes.
2478
+ */
2479
+ export declare class FunctionResponsePart {
2480
+ /** Optional. Inline media bytes. */
2481
+ inlineData?: FunctionResponseBlob;
2482
+ /** Optional. URI based data. */
2483
+ fileData?: FunctionResponseFileData;
2484
+ }
2485
+
2412
2486
  /** Specifies how the response should be scheduled in the conversation. */
2413
2487
  export declare enum FunctionResponseScheduling {
2414
2488
  /**
@@ -2592,6 +2666,9 @@ export declare interface GenerateContentConfig {
2592
2666
  /** The thinking features configuration.
2593
2667
  */
2594
2668
  thinkingConfig?: ThinkingConfig;
2669
+ /** The image generation configuration.
2670
+ */
2671
+ imageConfig?: ImageConfig;
2595
2672
  }
2596
2673
 
2597
2674
  /** Config for models.generate_content parameters. */
@@ -3725,6 +3802,13 @@ declare interface Image_2 {
3725
3802
  }
3726
3803
  export { Image_2 as Image }
3727
3804
 
3805
+ /** The image generation configuration to be used in GenerateContentConfig. */
3806
+ export declare interface ImageConfig {
3807
+ /** Aspect ratio of the generated images. Supported values are
3808
+ "1:1", "2:3", "3:2", "3:4", "4:3", "9:16", "16:9", and "21:9". */
3809
+ aspectRatio?: string;
3810
+ }
3811
+
3728
3812
  /** Enum that specifies the language of the text in the prompt. */
3729
3813
  export declare enum ImagePromptLanguage {
3730
3814
  /**
@@ -4304,6 +4388,11 @@ export declare interface LiveConnectConfig {
4304
4388
  /** The speech generation configuration.
4305
4389
  */
4306
4390
  speechConfig?: SpeechConfig;
4391
+ /** Config for thinking features.
4392
+ An error will be returned if this field is set for models that don't
4393
+ support thinking.
4394
+ */
4395
+ thinkingConfig?: ThinkingConfig;
4307
4396
  /** If enabled, the model will detect emotions and adapt its responses accordingly. */
4308
4397
  enableAffectiveDialog?: boolean;
4309
4398
  /** The user provided system instructions for the model.
@@ -5885,7 +5974,7 @@ export declare interface RecontextImageSource {
5885
5974
  productImages?: ProductImage[];
5886
5975
  }
5887
5976
 
5888
- export declare type ReferenceImage = RawReferenceImage | MaskReferenceImage | ControlReferenceImage | StyleReferenceImage | SubjectReferenceImage;
5977
+ export declare type ReferenceImage = RawReferenceImage | MaskReferenceImage | ControlReferenceImage | StyleReferenceImage | SubjectReferenceImage | ContentReferenceImage;
5889
5978
 
5890
5979
  /** Private class that represents a Reference image that is sent to API. */
5891
5980
  declare interface ReferenceImageAPIInternal {
@@ -6769,7 +6858,7 @@ export declare interface Tool {
6769
6858
  /** Optional. Tool to support the model interacting directly with the
6770
6859
  computer. If enabled, it automatically populates computer-use specific
6771
6860
  Function Declarations. */
6772
- computerUse?: ToolComputerUse;
6861
+ computerUse?: ComputerUse;
6773
6862
  /** Optional. CodeExecution tool type. Enables the model to execute code as part of generation. */
6774
6863
  codeExecution?: ToolCodeExecution;
6775
6864
  }
@@ -6778,12 +6867,6 @@ export declare interface Tool {
6778
6867
  export declare interface ToolCodeExecution {
6779
6868
  }
6780
6869
 
6781
- /** Tool to support computer use. */
6782
- export declare interface ToolComputerUse {
6783
- /** Required. The environment being operated. */
6784
- environment?: Environment;
6785
- }
6786
-
6787
6870
  /** Tool config.
6788
6871
 
6789
6872
  This config is shared for all tools provided in the request.
@@ -6938,6 +7021,8 @@ export declare interface TuningJob {
6938
7021
  serviceAccount?: string;
6939
7022
  /** Optional. The display name of the TunedModel. The name can be up to 128 characters long and can consist of any UTF-8 characters. */
6940
7023
  tunedModelDisplayName?: string;
7024
+ /** Tuning Spec for Veo Tuning. */
7025
+ veoTuningSpec?: VeoTuningSpec;
6941
7026
  }
6942
7027
 
6943
7028
  /** Tuning mode. */
@@ -7021,6 +7106,22 @@ declare class Tunings extends BaseModule {
7021
7106
  private tuneMldevInternal;
7022
7107
  }
7023
7108
 
7109
+ /** Optional. The tuning task. Either I2V or T2V. */
7110
+ export declare enum TuningTask {
7111
+ /**
7112
+ * Default value. This value is unused.
7113
+ */
7114
+ TUNING_TASK_UNSPECIFIED = "TUNING_TASK_UNSPECIFIED",
7115
+ /**
7116
+ * Tuning task for image to video.
7117
+ */
7118
+ TUNING_TASK_I2V = "TUNING_TASK_I2V",
7119
+ /**
7120
+ * Tuning task for text to video.
7121
+ */
7122
+ TUNING_TASK_T2V = "TUNING_TASK_T2V"
7123
+ }
7124
+
7024
7125
  export declare interface TuningValidationDataset {
7025
7126
  /** GCS URI of the file containing validation dataset in JSONL format. */
7026
7127
  gcsUri?: string;
@@ -7131,6 +7232,7 @@ declare namespace types {
7131
7232
  JobState,
7132
7233
  TuningMode,
7133
7234
  AdapterSize,
7235
+ TuningTask,
7134
7236
  FeatureSelectionPreference,
7135
7237
  Behavior,
7136
7238
  DynamicRetrievalConfigMode,
@@ -7165,6 +7267,9 @@ declare namespace types {
7165
7267
  FunctionCall,
7166
7268
  CodeExecutionResult,
7167
7269
  ExecutableCode,
7270
+ FunctionResponseBlob,
7271
+ FunctionResponseFileData,
7272
+ FunctionResponsePart,
7168
7273
  FunctionResponse,
7169
7274
  Part,
7170
7275
  Content,
@@ -7186,7 +7291,7 @@ declare namespace types {
7186
7291
  AuthConfig,
7187
7292
  GoogleMaps,
7188
7293
  UrlContext,
7189
- ToolComputerUse,
7294
+ ComputerUse,
7190
7295
  ApiAuthApiKeyConfig,
7191
7296
  ApiAuth,
7192
7297
  ExternalApiElasticSearchParams,
@@ -7216,6 +7321,7 @@ declare namespace types {
7216
7321
  SpeechConfig,
7217
7322
  AutomaticFunctionCallingConfig,
7218
7323
  ThinkingConfig,
7324
+ ImageConfig,
7219
7325
  GenerationConfigRoutingConfigAutoRoutingMode,
7220
7326
  GenerationConfigRoutingConfigManualRoutingMode,
7221
7327
  GenerationConfigRoutingConfig,
@@ -7338,6 +7444,8 @@ declare namespace types {
7338
7444
  TuningDataStats,
7339
7445
  EncryptionSpec,
7340
7446
  PartnerModelTuningSpec,
7447
+ VeoHyperParameters,
7448
+ VeoTuningSpec,
7341
7449
  TuningJob,
7342
7450
  ListTuningJobsConfig,
7343
7451
  ListTuningJobsParameters,
@@ -7421,6 +7529,7 @@ declare namespace types {
7421
7529
  ControlReferenceImage,
7422
7530
  StyleReferenceImage,
7423
7531
  SubjectReferenceImage,
7532
+ ContentReferenceImage,
7424
7533
  LiveServerSetupComplete,
7425
7534
  Transcription,
7426
7535
  LiveServerContent,
@@ -7714,6 +7823,26 @@ export declare interface UsageMetadata {
7714
7823
  trafficType?: TrafficType;
7715
7824
  }
7716
7825
 
7826
+ /** Hyperparameters for Veo. */
7827
+ export declare interface VeoHyperParameters {
7828
+ /** Optional. Number of complete passes the model makes over the entire training dataset during training. */
7829
+ epochCount?: string;
7830
+ /** Optional. Multiplier for adjusting the default learning rate. */
7831
+ learningRateMultiplier?: number;
7832
+ /** Optional. The tuning task. Either I2V or T2V. */
7833
+ tuningTask?: TuningTask;
7834
+ }
7835
+
7836
+ /** Tuning Spec for Veo Model Tuning. */
7837
+ export declare interface VeoTuningSpec {
7838
+ /** Optional. Hyperparameters for Veo. */
7839
+ hyperParameters?: VeoHyperParameters;
7840
+ /** Required. Training dataset used for tuning. The dataset can be specified as either a Cloud Storage path to a JSONL file or as the resource name of a Vertex Multimodal Dataset. */
7841
+ trainingDatasetUri?: string;
7842
+ /** Optional. Validation dataset used for tuning. The dataset can be specified as either a Cloud Storage path to a JSONL file or as the resource name of a Vertex Multimodal Dataset. */
7843
+ validationDatasetUri?: string;
7844
+ }
7845
+
7717
7846
  /** Retrieve from Vertex AI Search datastore or engine for grounding. datastore and engine are mutually exclusive. See https://cloud.google.com/products/agent-builder */
7718
7847
  export declare interface VertexAISearch {
7719
7848
  /** Specifications that define the specific DataStores to be searched, along with configurations for those data stores. This is only considered for Engines with multiple data stores. It should only be set if engine is used. */