@google/genai 0.14.1 → 0.15.0

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.
package/dist/web/web.d.ts CHANGED
@@ -1,3 +1,4 @@
1
+ import { Client } from '@modelcontextprotocol/sdk/client/index.js';
1
2
  import { GoogleAuthOptions } from 'google-auth-library';
2
3
 
3
4
  /** Marks the end of user activity.
@@ -10,8 +11,17 @@ export declare interface ActivityEnd {
10
11
 
11
12
  /** The different ways of handling user activity. */
12
13
  export declare enum ActivityHandling {
14
+ /**
15
+ * If unspecified, the default behavior is `START_OF_ACTIVITY_INTERRUPTS`.
16
+ */
13
17
  ACTIVITY_HANDLING_UNSPECIFIED = "ACTIVITY_HANDLING_UNSPECIFIED",
18
+ /**
19
+ * If true, start of activity will interrupt the model's response (also called "barge in"). The model's current response will be cut-off in the moment of the interruption. This is the default behavior.
20
+ */
14
21
  START_OF_ACTIVITY_INTERRUPTS = "START_OF_ACTIVITY_INTERRUPTS",
22
+ /**
23
+ * The model's response will not be interrupted.
24
+ */
15
25
  NO_INTERRUPTION = "NO_INTERRUPTION"
16
26
  }
17
27
 
@@ -25,12 +35,33 @@ export declare interface ActivityStart {
25
35
 
26
36
  /** Optional. Adapter size for tuning. */
27
37
  export declare enum AdapterSize {
38
+ /**
39
+ * Adapter size is unspecified.
40
+ */
28
41
  ADAPTER_SIZE_UNSPECIFIED = "ADAPTER_SIZE_UNSPECIFIED",
42
+ /**
43
+ * Adapter size 1.
44
+ */
29
45
  ADAPTER_SIZE_ONE = "ADAPTER_SIZE_ONE",
46
+ /**
47
+ * Adapter size 2.
48
+ */
30
49
  ADAPTER_SIZE_TWO = "ADAPTER_SIZE_TWO",
50
+ /**
51
+ * Adapter size 4.
52
+ */
31
53
  ADAPTER_SIZE_FOUR = "ADAPTER_SIZE_FOUR",
54
+ /**
55
+ * Adapter size 8.
56
+ */
32
57
  ADAPTER_SIZE_EIGHT = "ADAPTER_SIZE_EIGHT",
58
+ /**
59
+ * Adapter size 16.
60
+ */
33
61
  ADAPTER_SIZE_SIXTEEN = "ADAPTER_SIZE_SIXTEEN",
62
+ /**
63
+ * Adapter size 32.
64
+ */
34
65
  ADAPTER_SIZE_THIRTY_TWO = "ADAPTER_SIZE_THIRTY_TWO"
35
66
  }
36
67
 
@@ -166,6 +197,16 @@ export declare interface ApiKeyConfig {
166
197
  apiKeyString?: string;
167
198
  }
168
199
 
200
+ /** Representation of an audio chunk. */
201
+ export declare interface AudioChunk {
202
+ /** Raw byets of audio data. */
203
+ data?: string;
204
+ /** MIME type of the audio chunk. */
205
+ mimeType?: string;
206
+ /** Prompts and config used for generating this audio chunk. */
207
+ sourceMetadata?: LiveMusicSourceMetadata;
208
+ }
209
+
169
210
  /** The audio transcription configuration in Setup. */
170
211
  export declare interface AudioTranscriptionConfig {
171
212
  }
@@ -234,11 +275,29 @@ export declare interface AuthConfigOidcConfig {
234
275
  /** Type of auth scheme. */
235
276
  export declare enum AuthType {
236
277
  AUTH_TYPE_UNSPECIFIED = "AUTH_TYPE_UNSPECIFIED",
278
+ /**
279
+ * No Auth.
280
+ */
237
281
  NO_AUTH = "NO_AUTH",
282
+ /**
283
+ * API Key Auth.
284
+ */
238
285
  API_KEY_AUTH = "API_KEY_AUTH",
286
+ /**
287
+ * HTTP Basic Auth.
288
+ */
239
289
  HTTP_BASIC_AUTH = "HTTP_BASIC_AUTH",
290
+ /**
291
+ * Google Service Account Auth.
292
+ */
240
293
  GOOGLE_SERVICE_ACCOUNT_AUTH = "GOOGLE_SERVICE_ACCOUNT_AUTH",
294
+ /**
295
+ * OAuth auth.
296
+ */
241
297
  OAUTH = "OAUTH",
298
+ /**
299
+ * OpenID Connect (OIDC) Auth.
300
+ */
242
301
  OIDC_AUTH = "OIDC_AUTH"
243
302
  }
244
303
 
@@ -256,6 +315,28 @@ export declare interface AutomaticActivityDetection {
256
315
  silenceDurationMs?: number;
257
316
  }
258
317
 
318
+ /** The configuration for automatic function calling. */
319
+ export declare interface AutomaticFunctionCallingConfig {
320
+ /** Whether to disable automatic function calling.
321
+ If not set or set to False, will enable automatic function calling.
322
+ If set to True, will disable automatic function calling.
323
+ */
324
+ disable?: boolean;
325
+ /** If automatic function calling is enabled,
326
+ maximum number of remote calls for automatic function calling.
327
+ This number should be a positive integer.
328
+ If not set, SDK will set maximum number of remote calls to 10.
329
+ */
330
+ maximumRemoteCalls?: number;
331
+ /** If automatic function calling is enabled,
332
+ whether to ignore call history to the response.
333
+ If not set, SDK will set ignore_call_history to false,
334
+ and will append the call history to
335
+ GenerateContentResponse.automatic_function_calling_history.
336
+ */
337
+ ignoreCallHistory?: boolean;
338
+ }
339
+
259
340
  /**
260
341
  * @license
261
342
  * Copyright 2025 Google LLC
@@ -272,6 +353,22 @@ export declare interface BaseUrlParameters {
272
353
  vertexUrl?: string;
273
354
  }
274
355
 
356
+ /** Defines the function behavior. Defaults to `BLOCKING`. */
357
+ export declare enum Behavior {
358
+ /**
359
+ * This value is unused.
360
+ */
361
+ UNSPECIFIED = "UNSPECIFIED",
362
+ /**
363
+ * If set, the system will wait to receive the function response before continuing the conversation.
364
+ */
365
+ BLOCKING = "BLOCKING",
366
+ /**
367
+ * If set, the system will not wait to receive the function response. Instead, it will attempt to handle function responses as they become available while maintaining the conversation between the user and the model.
368
+ */
369
+ NON_BLOCKING = "NON_BLOCKING"
370
+ }
371
+
275
372
  /** Content blob. */
276
373
  declare interface Blob_2 {
277
374
  /** Optional. Display name of the blob. Used to provide a label or filename to distinguish blobs. This field is not currently used in the Gemini GenerateContent calls. */
@@ -287,10 +384,25 @@ export declare type BlobImageUnion = Blob_2;
287
384
 
288
385
  /** Output only. Blocked reason. */
289
386
  export declare enum BlockedReason {
387
+ /**
388
+ * Unspecified blocked reason.
389
+ */
290
390
  BLOCKED_REASON_UNSPECIFIED = "BLOCKED_REASON_UNSPECIFIED",
391
+ /**
392
+ * Candidates blocked due to safety.
393
+ */
291
394
  SAFETY = "SAFETY",
395
+ /**
396
+ * Candidates blocked due to other reason.
397
+ */
292
398
  OTHER = "OTHER",
399
+ /**
400
+ * Candidates blocked due to the terms which are included from the terminology blocklist.
401
+ */
293
402
  BLOCKLIST = "BLOCKLIST",
403
+ /**
404
+ * Candidates blocked due to prohibited content.
405
+ */
294
406
  PROHIBITED_CONTENT = "PROHIBITED_CONTENT"
295
407
  }
296
408
 
@@ -413,6 +525,33 @@ export declare class Caches extends BaseModule {
413
525
  private listInternal;
414
526
  }
415
527
 
528
+ /**
529
+ * CallableTool is an invokable tool that can be executed with external
530
+ * application (e.g., via Model Context Protocol) or local functions with
531
+ * function calling.
532
+ */
533
+ export declare interface CallableTool {
534
+ /**
535
+ * Returns tool that can be called by Gemini.
536
+ */
537
+ tool(): Promise<Tool>;
538
+ /**
539
+ * Executes the callable tool with the given function call arguments and
540
+ * returns the response parts from the tool execution.
541
+ */
542
+ callTool(functionCalls: FunctionCall[]): Promise<Part[]>;
543
+ }
544
+
545
+ /**
546
+ * CallableToolConfig is the configuration for a callable tool.
547
+ */
548
+ export declare interface CallableToolConfig {
549
+ /**
550
+ * Specifies the model's behavior after invoking this tool.
551
+ */
552
+ behavior?: Behavior;
553
+ }
554
+
416
555
  /** A response candidate generated from the model. */
417
556
  export declare interface Candidate {
418
557
  /** Contains the multi-part content of the response.
@@ -431,6 +570,8 @@ export declare interface Candidate {
431
570
  If empty, the model has not stopped generating the tokens.
432
571
  */
433
572
  finishReason?: FinishReason;
573
+ /** Metadata related to url context retrieval tool. */
574
+ urlContextMetadata?: UrlContextMetadata;
434
575
  /** Output only. Average log probability score of the candidate. */
435
576
  avgLogprobs?: number;
436
577
  /** Output only. Metadata specifies sources used to ground generated content. */
@@ -798,6 +939,16 @@ export declare interface CreateCachedContentConfig {
798
939
  /** Configuration for the tools to use. This config is shared for all tools.
799
940
  */
800
941
  toolConfig?: ToolConfig;
942
+ /** The Cloud KMS resource identifier of the customer managed
943
+ encryption key used to protect a resource.
944
+ The key needs to be in the same region as where the compute resource is
945
+ created. See
946
+ https://cloud.google.com/vertex-ai/docs/general/cmek for more
947
+ details. If this is set, then all created CachedContent objects
948
+ will be encrypted with the provided encryption key.
949
+ Allowed formats: projects/{project}/locations/{location}/keyRings/{key_ring}/cryptoKeys/{crypto_key}
950
+ */
951
+ kmsKeyName?: string;
801
952
  }
802
953
 
803
954
  /** Parameters for caches.create method. */
@@ -1159,7 +1310,13 @@ export declare interface DynamicRetrievalConfig {
1159
1310
 
1160
1311
  /** Config for the dynamic retrieval config mode. */
1161
1312
  export declare enum DynamicRetrievalConfigMode {
1313
+ /**
1314
+ * Always trigger retrieval.
1315
+ */
1162
1316
  MODE_UNSPECIFIED = "MODE_UNSPECIFIED",
1317
+ /**
1318
+ * Run retrieval only when system decides it is necessary.
1319
+ */
1163
1320
  MODE_DYNAMIC = "MODE_DYNAMIC"
1164
1321
  }
1165
1322
 
@@ -1338,8 +1495,17 @@ export declare interface Endpoint {
1338
1495
 
1339
1496
  /** End of speech sensitivity. */
1340
1497
  export declare enum EndSensitivity {
1498
+ /**
1499
+ * The default is END_SENSITIVITY_LOW.
1500
+ */
1341
1501
  END_SENSITIVITY_UNSPECIFIED = "END_SENSITIVITY_UNSPECIFIED",
1502
+ /**
1503
+ * Automatic detection ends speech more often.
1504
+ */
1342
1505
  END_SENSITIVITY_HIGH = "END_SENSITIVITY_HIGH",
1506
+ /**
1507
+ * Automatic detection ends speech less often.
1508
+ */
1343
1509
  END_SENSITIVITY_LOW = "END_SENSITIVITY_LOW"
1344
1510
  }
1345
1511
 
@@ -1588,17 +1754,53 @@ export declare interface FileStatus {
1588
1754
  If empty, the model has not stopped generating the tokens.
1589
1755
  */
1590
1756
  export declare enum FinishReason {
1757
+ /**
1758
+ * The finish reason is unspecified.
1759
+ */
1591
1760
  FINISH_REASON_UNSPECIFIED = "FINISH_REASON_UNSPECIFIED",
1761
+ /**
1762
+ * Token generation reached a natural stopping point or a configured stop sequence.
1763
+ */
1592
1764
  STOP = "STOP",
1765
+ /**
1766
+ * Token generation reached the configured maximum output tokens.
1767
+ */
1593
1768
  MAX_TOKENS = "MAX_TOKENS",
1769
+ /**
1770
+ * Token generation stopped because the content potentially contains safety violations. NOTE: When streaming, [content][] is empty if content filters blocks the output.
1771
+ */
1594
1772
  SAFETY = "SAFETY",
1773
+ /**
1774
+ * The token generation stopped because of potential recitation.
1775
+ */
1595
1776
  RECITATION = "RECITATION",
1777
+ /**
1778
+ * The token generation stopped because of using an unsupported language.
1779
+ */
1596
1780
  LANGUAGE = "LANGUAGE",
1781
+ /**
1782
+ * All other reasons that stopped the token generation.
1783
+ */
1597
1784
  OTHER = "OTHER",
1785
+ /**
1786
+ * Token generation stopped because the content contains forbidden terms.
1787
+ */
1598
1788
  BLOCKLIST = "BLOCKLIST",
1789
+ /**
1790
+ * Token generation stopped for potentially containing prohibited content.
1791
+ */
1599
1792
  PROHIBITED_CONTENT = "PROHIBITED_CONTENT",
1793
+ /**
1794
+ * Token generation stopped because the content potentially contains Sensitive Personally Identifiable Information (SPII).
1795
+ */
1600
1796
  SPII = "SPII",
1797
+ /**
1798
+ * The function call generated by the model is invalid.
1799
+ */
1601
1800
  MALFORMED_FUNCTION_CALL = "MALFORMED_FUNCTION_CALL",
1801
+ /**
1802
+ * Token generation stopped because generated images have safety violations.
1803
+ */
1602
1804
  IMAGE_SAFETY = "IMAGE_SAFETY"
1603
1805
  }
1604
1806
 
@@ -1623,14 +1825,32 @@ export declare interface FunctionCallingConfig {
1623
1825
 
1624
1826
  /** Config for the function calling config mode. */
1625
1827
  export declare enum FunctionCallingConfigMode {
1828
+ /**
1829
+ * The function calling config mode is unspecified. Should not be used.
1830
+ */
1626
1831
  MODE_UNSPECIFIED = "MODE_UNSPECIFIED",
1832
+ /**
1833
+ * Default model behavior, model decides to predict either function calls or natural language response.
1834
+ */
1627
1835
  AUTO = "AUTO",
1836
+ /**
1837
+ * Model is constrained to always predicting function calls only. 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".
1838
+ */
1628
1839
  ANY = "ANY",
1840
+ /**
1841
+ * Model will not predict any function calls. Model behavior is same as when not passing any function declarations.
1842
+ */
1629
1843
  NONE = "NONE"
1630
1844
  }
1631
1845
 
1632
- /** 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. */
1846
+ /** Defines a function that the model can generate JSON inputs for.
1847
+
1848
+ The inputs are based on `OpenAPI 3.0 specifications
1849
+ <https://spec.openapis.org/oas/v3.0.3>`_.
1850
+ */
1633
1851
  export declare interface FunctionDeclaration {
1852
+ /** Defines the function behavior. */
1853
+ behavior?: Behavior;
1634
1854
  /** Optional. Description and purpose of the function. Model uses it to decide how and whether to call the function. */
1635
1855
  description?: string;
1636
1856
  /** 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. */
@@ -1643,8 +1863,11 @@ export declare interface FunctionDeclaration {
1643
1863
 
1644
1864
  /** A function response. */
1645
1865
  export declare class FunctionResponse {
1646
- /** The id of the function call this response is for. Populated by the client
1647
- to match the corresponding function call `id`. */
1866
+ /** Signals that function call continues, and more responses will be returned, turning the function call into a generator. Is only applicable to NON_BLOCKING function calls (see FunctionDeclaration.behavior for details), ignored otherwise. If false, the default, future responses will not be considered. Is only applicable to NON_BLOCKING function calls, is ignored otherwise. If set to false, future responses will not be considered. It is allowed to return empty `response` with `will_continue=False` to signal that the function call is finished. */
1867
+ willContinue?: boolean;
1868
+ /** Specifies how the response should be scheduled in the conversation. Only applicable to NON_BLOCKING function calls, is ignored otherwise. Defaults to WHEN_IDLE. */
1869
+ scheduling?: FunctionResponseScheduling;
1870
+ /** Optional. The id of the function call this response is for. Populated by the client to match the corresponding function call `id`. */
1648
1871
  id?: string;
1649
1872
  /** Required. The name of the function to call. Matches [FunctionDeclaration.name] and [FunctionCall.name]. */
1650
1873
  name?: string;
@@ -1652,6 +1875,26 @@ export declare class FunctionResponse {
1652
1875
  response?: Record<string, unknown>;
1653
1876
  }
1654
1877
 
1878
+ /** Specifies how the response should be scheduled in the conversation. */
1879
+ export declare enum FunctionResponseScheduling {
1880
+ /**
1881
+ * This value is unused.
1882
+ */
1883
+ SCHEDULING_UNSPECIFIED = "SCHEDULING_UNSPECIFIED",
1884
+ /**
1885
+ * Only add the result to the conversation context, do not interrupt or trigger generation.
1886
+ */
1887
+ SILENT = "SILENT",
1888
+ /**
1889
+ * Add the result to the conversation context, and prompt to generate output without interrupting ongoing generation.
1890
+ */
1891
+ WHEN_IDLE = "WHEN_IDLE",
1892
+ /**
1893
+ * Add the result to the conversation context, interrupt ongoing generation and prompt to generate output.
1894
+ */
1895
+ INTERRUPT = "INTERRUPT"
1896
+ }
1897
+
1655
1898
  /** Optional model configuration parameters.
1656
1899
 
1657
1900
  For more information, see `Content generation parameters
@@ -1777,6 +2020,9 @@ export declare interface GenerateContentConfig {
1777
2020
  model.
1778
2021
  */
1779
2022
  audioTimestamp?: boolean;
2023
+ /** The configuration for automatic function calling.
2024
+ */
2025
+ automaticFunctionCalling?: AutomaticFunctionCallingConfig;
1780
2026
  /** The thinking features configuration.
1781
2027
  */
1782
2028
  thinkingConfig?: ThinkingConfig;
@@ -1806,6 +2052,9 @@ export declare class GenerateContentResponse {
1806
2052
  /** Identifier for each response.
1807
2053
  */
1808
2054
  responseId?: string;
2055
+ /** The history of automatic function calling.
2056
+ */
2057
+ automaticFunctionCallingHistory?: Content[];
1809
2058
  /** Output only. The model version used to generate the response. */
1810
2059
  modelVersion?: string;
1811
2060
  /** Output only. Content filter results for a prompt sent in the request. Note: Sent only in the first stream chunk. Only happens when no candidates were generated due to content violations. */
@@ -2454,6 +2703,10 @@ export declare interface GoogleRpcStatus {
2454
2703
 
2455
2704
  /** Tool to support Google Search in Model. Powered by Google. */
2456
2705
  export declare interface GoogleSearch {
2706
+ /** Optional. Filter search results to a specific time range.
2707
+ If customers set a start time, they must set an end time (and vice versa).
2708
+ */
2709
+ timeRangeFilter?: Interval;
2457
2710
  }
2458
2711
 
2459
2712
  /** Tool to retrieve public web data for grounding, powered by Google. */
@@ -2528,46 +2781,121 @@ export declare interface GroundingSupport {
2528
2781
 
2529
2782
  /** Optional. Specify if the threshold is used for probability or severity score. If not specified, the threshold is used for probability score. */
2530
2783
  export declare enum HarmBlockMethod {
2784
+ /**
2785
+ * The harm block method is unspecified.
2786
+ */
2531
2787
  HARM_BLOCK_METHOD_UNSPECIFIED = "HARM_BLOCK_METHOD_UNSPECIFIED",
2788
+ /**
2789
+ * The harm block method uses both probability and severity scores.
2790
+ */
2532
2791
  SEVERITY = "SEVERITY",
2792
+ /**
2793
+ * The harm block method uses the probability score.
2794
+ */
2533
2795
  PROBABILITY = "PROBABILITY"
2534
2796
  }
2535
2797
 
2536
2798
  /** Required. The harm block threshold. */
2537
2799
  export declare enum HarmBlockThreshold {
2800
+ /**
2801
+ * Unspecified harm block threshold.
2802
+ */
2538
2803
  HARM_BLOCK_THRESHOLD_UNSPECIFIED = "HARM_BLOCK_THRESHOLD_UNSPECIFIED",
2804
+ /**
2805
+ * Block low threshold and above (i.e. block more).
2806
+ */
2539
2807
  BLOCK_LOW_AND_ABOVE = "BLOCK_LOW_AND_ABOVE",
2808
+ /**
2809
+ * Block medium threshold and above.
2810
+ */
2540
2811
  BLOCK_MEDIUM_AND_ABOVE = "BLOCK_MEDIUM_AND_ABOVE",
2812
+ /**
2813
+ * Block only high threshold (i.e. block less).
2814
+ */
2541
2815
  BLOCK_ONLY_HIGH = "BLOCK_ONLY_HIGH",
2816
+ /**
2817
+ * Block none.
2818
+ */
2542
2819
  BLOCK_NONE = "BLOCK_NONE",
2820
+ /**
2821
+ * Turn off the safety filter.
2822
+ */
2543
2823
  OFF = "OFF"
2544
2824
  }
2545
2825
 
2546
2826
  /** Required. Harm category. */
2547
2827
  export declare enum HarmCategory {
2828
+ /**
2829
+ * The harm category is unspecified.
2830
+ */
2548
2831
  HARM_CATEGORY_UNSPECIFIED = "HARM_CATEGORY_UNSPECIFIED",
2832
+ /**
2833
+ * The harm category is hate speech.
2834
+ */
2549
2835
  HARM_CATEGORY_HATE_SPEECH = "HARM_CATEGORY_HATE_SPEECH",
2836
+ /**
2837
+ * The harm category is dangerous content.
2838
+ */
2550
2839
  HARM_CATEGORY_DANGEROUS_CONTENT = "HARM_CATEGORY_DANGEROUS_CONTENT",
2840
+ /**
2841
+ * The harm category is harassment.
2842
+ */
2551
2843
  HARM_CATEGORY_HARASSMENT = "HARM_CATEGORY_HARASSMENT",
2844
+ /**
2845
+ * The harm category is sexually explicit content.
2846
+ */
2552
2847
  HARM_CATEGORY_SEXUALLY_EXPLICIT = "HARM_CATEGORY_SEXUALLY_EXPLICIT",
2848
+ /**
2849
+ * The harm category is civic integrity.
2850
+ */
2553
2851
  HARM_CATEGORY_CIVIC_INTEGRITY = "HARM_CATEGORY_CIVIC_INTEGRITY"
2554
2852
  }
2555
2853
 
2556
2854
  /** Output only. Harm probability levels in the content. */
2557
2855
  export declare enum HarmProbability {
2856
+ /**
2857
+ * Harm probability unspecified.
2858
+ */
2558
2859
  HARM_PROBABILITY_UNSPECIFIED = "HARM_PROBABILITY_UNSPECIFIED",
2860
+ /**
2861
+ * Negligible level of harm.
2862
+ */
2559
2863
  NEGLIGIBLE = "NEGLIGIBLE",
2864
+ /**
2865
+ * Low level of harm.
2866
+ */
2560
2867
  LOW = "LOW",
2868
+ /**
2869
+ * Medium level of harm.
2870
+ */
2561
2871
  MEDIUM = "MEDIUM",
2872
+ /**
2873
+ * High level of harm.
2874
+ */
2562
2875
  HIGH = "HIGH"
2563
2876
  }
2564
2877
 
2565
2878
  /** Output only. Harm severity levels in the content. */
2566
2879
  export declare enum HarmSeverity {
2880
+ /**
2881
+ * Harm severity unspecified.
2882
+ */
2567
2883
  HARM_SEVERITY_UNSPECIFIED = "HARM_SEVERITY_UNSPECIFIED",
2884
+ /**
2885
+ * Negligible level of harm severity.
2886
+ */
2568
2887
  HARM_SEVERITY_NEGLIGIBLE = "HARM_SEVERITY_NEGLIGIBLE",
2888
+ /**
2889
+ * Low level of harm severity.
2890
+ */
2569
2891
  HARM_SEVERITY_LOW = "HARM_SEVERITY_LOW",
2892
+ /**
2893
+ * Medium level of harm severity.
2894
+ */
2570
2895
  HARM_SEVERITY_MEDIUM = "HARM_SEVERITY_MEDIUM",
2896
+ /**
2897
+ * High level of harm severity.
2898
+ */
2571
2899
  HARM_SEVERITY_HIGH = "HARM_SEVERITY_HIGH"
2572
2900
  }
2573
2901
 
@@ -2668,25 +2996,81 @@ export declare enum ImagePromptLanguage {
2668
2996
  hi = "hi"
2669
2997
  }
2670
2998
 
2999
+ /** Represents a time interval, encoded as a start time (inclusive) and an end time (exclusive).
3000
+
3001
+ The start time must be less than or equal to the end time.
3002
+ When the start equals the end time, the interval is an empty interval.
3003
+ (matches no time)
3004
+ When both start and end are unspecified, the interval matches any time.
3005
+ */
3006
+ export declare interface Interval {
3007
+ /** The start time of the interval. */
3008
+ startTime?: string;
3009
+ /** The end time of the interval. */
3010
+ endTime?: string;
3011
+ }
3012
+
2671
3013
  /** Output only. The detailed state of the job. */
2672
3014
  export declare enum JobState {
3015
+ /**
3016
+ * The job state is unspecified.
3017
+ */
2673
3018
  JOB_STATE_UNSPECIFIED = "JOB_STATE_UNSPECIFIED",
3019
+ /**
3020
+ * The job has been just created or resumed and processing has not yet begun.
3021
+ */
2674
3022
  JOB_STATE_QUEUED = "JOB_STATE_QUEUED",
3023
+ /**
3024
+ * The service is preparing to run the job.
3025
+ */
2675
3026
  JOB_STATE_PENDING = "JOB_STATE_PENDING",
3027
+ /**
3028
+ * The job is in progress.
3029
+ */
2676
3030
  JOB_STATE_RUNNING = "JOB_STATE_RUNNING",
3031
+ /**
3032
+ * The job completed successfully.
3033
+ */
2677
3034
  JOB_STATE_SUCCEEDED = "JOB_STATE_SUCCEEDED",
3035
+ /**
3036
+ * The job failed.
3037
+ */
2678
3038
  JOB_STATE_FAILED = "JOB_STATE_FAILED",
3039
+ /**
3040
+ * The job is being cancelled. From this state the job may only go to either `JOB_STATE_SUCCEEDED`, `JOB_STATE_FAILED` or `JOB_STATE_CANCELLED`.
3041
+ */
2679
3042
  JOB_STATE_CANCELLING = "JOB_STATE_CANCELLING",
3043
+ /**
3044
+ * The job has been cancelled.
3045
+ */
2680
3046
  JOB_STATE_CANCELLED = "JOB_STATE_CANCELLED",
3047
+ /**
3048
+ * The job has been stopped, and can be resumed.
3049
+ */
2681
3050
  JOB_STATE_PAUSED = "JOB_STATE_PAUSED",
3051
+ /**
3052
+ * The job has expired.
3053
+ */
2682
3054
  JOB_STATE_EXPIRED = "JOB_STATE_EXPIRED",
3055
+ /**
3056
+ * The job is being updated. Only jobs in the `RUNNING` state can be updated. After updating, the job goes back to the `RUNNING` state.
3057
+ */
2683
3058
  JOB_STATE_UPDATING = "JOB_STATE_UPDATING",
3059
+ /**
3060
+ * The job is partially succeeded, some results may be missing due to errors.
3061
+ */
2684
3062
  JOB_STATE_PARTIALLY_SUCCEEDED = "JOB_STATE_PARTIALLY_SUCCEEDED"
2685
3063
  }
2686
3064
 
2687
3065
  /** Required. Programming language of the `code`. */
2688
3066
  export declare enum Language {
3067
+ /**
3068
+ * Unspecified language. This value should not be used.
3069
+ */
2689
3070
  LANGUAGE_UNSPECIFIED = "LANGUAGE_UNSPECIFIED",
3071
+ /**
3072
+ * Python >= 3.10, with numpy and simpy available.
3073
+ */
2690
3074
  PYTHON = "PYTHON"
2691
3075
  }
2692
3076
 
@@ -2827,12 +3211,14 @@ export declare class Live {
2827
3211
  private readonly apiClient;
2828
3212
  private readonly auth;
2829
3213
  private readonly webSocketFactory;
3214
+ readonly music: LiveMusic;
2830
3215
  constructor(apiClient: ApiClient, auth: Auth, webSocketFactory: WebSocketFactory);
2831
3216
  /**
2832
3217
  Establishes a connection to the specified model with the given
2833
3218
  configuration and returns a Session object representing that connection.
2834
3219
 
2835
- @experimental
3220
+ @experimental Built-in MCP support is a preview feature, may change in
3221
+ future versions.
2836
3222
 
2837
3223
  @remarks
2838
3224
 
@@ -2870,6 +3256,7 @@ export declare class Live {
2870
3256
  ```
2871
3257
  */
2872
3258
  connect(params: types.LiveConnectParameters): Promise<Session>;
3259
+ private isCallableTool;
2873
3260
  }
2874
3261
 
2875
3262
  /** Callbacks for the live API. */
@@ -3003,6 +3390,9 @@ export declare interface LiveClientSetup {
3003
3390
  specified for the output audio.
3004
3391
  */
3005
3392
  outputAudioTranscription?: AudioTranscriptionConfig;
3393
+ /** Configures the proactivity of the model. This allows the model to respond proactively to
3394
+ the input and to ignore irrelevant input. */
3395
+ proactivity?: ProactivityConfig;
3006
3396
  }
3007
3397
 
3008
3398
  /** Client generated response to a `ToolCall` received from the server.
@@ -3060,6 +3450,8 @@ export declare interface LiveConnectConfig {
3060
3450
  /** The speech generation configuration.
3061
3451
  */
3062
3452
  speechConfig?: SpeechConfig;
3453
+ /** If enabled, the model will detect emotions and adapt its responses accordingly. */
3454
+ enableAffectiveDialog?: boolean;
3063
3455
  /** The user provided system instructions for the model.
3064
3456
  Note: only text should be used in parts and content in each part will be
3065
3457
  in a separate paragraph. */
@@ -3087,6 +3479,9 @@ export declare interface LiveConnectConfig {
3087
3479
 
3088
3480
  If included, server will compress context window to fit into given length. */
3089
3481
  contextWindowCompression?: ContextWindowCompressionConfig;
3482
+ /** Configures the proactivity of the model. This allows the model to respond proactively to
3483
+ the input and to ignore irrelevant input. */
3484
+ proactivity?: ProactivityConfig;
3090
3485
  }
3091
3486
 
3092
3487
  /** Parameters for connecting to the live API. */
@@ -3101,6 +3496,288 @@ export declare interface LiveConnectParameters {
3101
3496
  config?: LiveConnectConfig;
3102
3497
  }
3103
3498
 
3499
+ /**
3500
+ LiveMusic class encapsulates the configuration for live music
3501
+ generation via Lyria Live models.
3502
+
3503
+ @experimental
3504
+ */
3505
+ declare class LiveMusic {
3506
+ private readonly apiClient;
3507
+ private readonly auth;
3508
+ private readonly webSocketFactory;
3509
+ constructor(apiClient: ApiClient, auth: Auth, webSocketFactory: WebSocketFactory);
3510
+ /**
3511
+ Establishes a connection to the specified model and returns a
3512
+ LiveMusicSession object representing that connection.
3513
+
3514
+ @experimental
3515
+
3516
+ @remarks
3517
+
3518
+ @param params - The parameters for establishing a connection to the model.
3519
+ @return A live session.
3520
+
3521
+ @example
3522
+ ```ts
3523
+ let model = 'models/lyria-realtime-exp';
3524
+ const session = await ai.live.music.connect({
3525
+ model: model,
3526
+ callbacks: {
3527
+ onmessage: (e: MessageEvent) => {
3528
+ console.log('Received message from the server: %s\n', debug(e.data));
3529
+ },
3530
+ onerror: (e: ErrorEvent) => {
3531
+ console.log('Error occurred: %s\n', debug(e.error));
3532
+ },
3533
+ onclose: (e: CloseEvent) => {
3534
+ console.log('Connection closed.');
3535
+ },
3536
+ },
3537
+ });
3538
+ ```
3539
+ */
3540
+ connect(params: types.LiveMusicConnectParameters): Promise<LiveMusicSession>;
3541
+ }
3542
+
3543
+ /** Callbacks for the realtime music API. */
3544
+ export declare interface LiveMusicCallbacks {
3545
+ /**
3546
+ * Called when a message is received from the server.
3547
+ */
3548
+ onmessage: (e: LiveMusicServerMessage) => void;
3549
+ /**
3550
+ * Called when an error occurs.
3551
+ */
3552
+ onerror?: ((e: ErrorEvent) => void) | null;
3553
+ /**
3554
+ * Called when the websocket connection is closed.
3555
+ */
3556
+ onclose?: ((e: CloseEvent) => void) | null;
3557
+ }
3558
+
3559
+ /** User input to start or steer the music. */
3560
+ export declare interface LiveMusicClientContent {
3561
+ /** Weighted prompts as the model input. */
3562
+ weightedPrompts?: WeightedPrompt[];
3563
+ }
3564
+
3565
+ /** Messages sent by the client in the LiveMusicClientMessage call. */
3566
+ export declare interface LiveMusicClientMessage {
3567
+ /** Message to be sent in the first (and only in the first) `LiveMusicClientMessage`.
3568
+ Clients should wait for a `LiveMusicSetupComplete` message before
3569
+ sending any additional messages. */
3570
+ setup?: LiveMusicClientSetup;
3571
+ /** User input to influence music generation. */
3572
+ clientContent?: LiveMusicClientContent;
3573
+ /** Configuration for music generation. */
3574
+ musicGenerationConfig?: LiveMusicGenerationConfig;
3575
+ /** Playback control signal for the music generation. */
3576
+ playbackControl?: LiveMusicPlaybackControl;
3577
+ }
3578
+
3579
+ /** Message to be sent by the system when connecting to the API. */
3580
+ export declare interface LiveMusicClientSetup {
3581
+ /** The model's resource name. Format: `models/{model}`. */
3582
+ model?: string;
3583
+ }
3584
+
3585
+ /** Parameters for connecting to the live API. */
3586
+ export declare interface LiveMusicConnectParameters {
3587
+ /** The model's resource name. */
3588
+ model: string;
3589
+ /** Callbacks invoked on server events. */
3590
+ callbacks: LiveMusicCallbacks;
3591
+ }
3592
+
3593
+ /** A prompt that was filtered with the reason. */
3594
+ export declare interface LiveMusicFilteredPrompt {
3595
+ /** The text prompt that was filtered. */
3596
+ text?: string;
3597
+ /** The reason the prompt was filtered. */
3598
+ filteredReason?: string;
3599
+ }
3600
+
3601
+ /** Configuration for music generation. */
3602
+ export declare interface LiveMusicGenerationConfig {
3603
+ /** Controls the variance in audio generation. Higher values produce
3604
+ higher variance. Range is [0.0, 3.0]. */
3605
+ temperature?: number;
3606
+ /** Controls how the model selects tokens for output. Samples the topK
3607
+ tokens with the highest probabilities. Range is [1, 1000]. */
3608
+ topK?: number;
3609
+ /** Seeds audio generation. If not set, the request uses a randomly
3610
+ generated seed. */
3611
+ seed?: number;
3612
+ /** Controls how closely the model follows prompts.
3613
+ Higher guidance follows more closely, but will make transitions more
3614
+ abrupt. Range is [0.0, 6.0]. */
3615
+ guidance?: number;
3616
+ /** Beats per minute. Range is [60, 200]. */
3617
+ bpm?: number;
3618
+ /** Density of sounds. Range is [0.0, 1.0]. */
3619
+ density?: number;
3620
+ /** Brightness of the music. Range is [0.0, 1.0]. */
3621
+ brightness?: number;
3622
+ /** Scale of the generated music. */
3623
+ scale?: Scale;
3624
+ /** Whether the audio output should contain bass. */
3625
+ muteBass?: boolean;
3626
+ /** Whether the audio output should contain drums. */
3627
+ muteDrums?: boolean;
3628
+ /** Whether the audio output should contain only bass and drums. */
3629
+ onlyBassAndDrums?: boolean;
3630
+ /** The mode of music generation. Default mode is QUALITY. */
3631
+ musicGenerationMode?: MusicGenerationMode;
3632
+ }
3633
+
3634
+ /** The playback control signal to apply to the music generation. */
3635
+ export declare enum LiveMusicPlaybackControl {
3636
+ /**
3637
+ * This value is unused.
3638
+ */
3639
+ PLAYBACK_CONTROL_UNSPECIFIED = "PLAYBACK_CONTROL_UNSPECIFIED",
3640
+ /**
3641
+ * Start generating the music.
3642
+ */
3643
+ PLAY = "PLAY",
3644
+ /**
3645
+ * Hold the music generation. Use PLAY to resume from the current position.
3646
+ */
3647
+ PAUSE = "PAUSE",
3648
+ /**
3649
+ * Stop the music generation and reset the context (prompts retained).
3650
+ Use PLAY to restart the music generation.
3651
+ */
3652
+ STOP = "STOP",
3653
+ /**
3654
+ * Reset the context of the music generation without stopping it.
3655
+ Retains the current prompts and config.
3656
+ */
3657
+ RESET_CONTEXT = "RESET_CONTEXT"
3658
+ }
3659
+
3660
+ /** Server update generated by the model in response to client messages.
3661
+
3662
+ Content is generated as quickly as possible, and not in real time.
3663
+ Clients may choose to buffer and play it out in real time.
3664
+ */
3665
+ export declare interface LiveMusicServerContent {
3666
+ /** The audio chunks that the model has generated. */
3667
+ audioChunks?: AudioChunk[];
3668
+ }
3669
+
3670
+ /** Response message for the LiveMusicClientMessage call. */
3671
+ export declare class LiveMusicServerMessage {
3672
+ /** Message sent in response to a `LiveMusicClientSetup` message from the client.
3673
+ Clients should wait for this message before sending any additional messages. */
3674
+ setupComplete?: LiveMusicServerSetupComplete;
3675
+ /** Content generated by the model in response to client messages. */
3676
+ serverContent?: LiveMusicServerContent;
3677
+ /** A prompt that was filtered with the reason. */
3678
+ filteredPrompt?: LiveMusicFilteredPrompt;
3679
+ /**
3680
+ * Returns the first audio chunk from the server content, if present.
3681
+ *
3682
+ * @remarks
3683
+ * If there are no audio chunks in the response, undefined will be returned.
3684
+ */
3685
+ get audioChunk(): AudioChunk | undefined;
3686
+ }
3687
+
3688
+ /** Sent in response to a `LiveMusicClientSetup` message from the client. */
3689
+ export declare interface LiveMusicServerSetupComplete {
3690
+ }
3691
+
3692
+ /**
3693
+ Represents a connection to the API.
3694
+
3695
+ @experimental
3696
+ */
3697
+ declare class LiveMusicSession {
3698
+ readonly conn: WebSocket_2;
3699
+ private readonly apiClient;
3700
+ constructor(conn: WebSocket_2, apiClient: ApiClient);
3701
+ /**
3702
+ Sets inputs to steer music generation. Updates the session's current
3703
+ weighted prompts.
3704
+
3705
+ @param params - Contains one property, `weightedPrompts`.
3706
+
3707
+ - `weightedPrompts` to send to the model; weights are normalized to
3708
+ sum to 1.0.
3709
+
3710
+ @experimental
3711
+ */
3712
+ setClientContent(params: types.LiveMusicSetClientContentParameters): Promise<void>;
3713
+ /**
3714
+ Sets a configuration to the model. Updates the session's current
3715
+ music generation config.
3716
+
3717
+ @param params - Contains one property, `musicGenerationConfig`.
3718
+
3719
+ - `musicGenerationConfig` to set in the model. Passing an empty or
3720
+ undefined config to the model will reset the config to defaults.
3721
+
3722
+ @experimental
3723
+ */
3724
+ setMusicGenerationConfig(params: types.LiveMusicSetConfigParameters): Promise<void>;
3725
+ private sendPlaybackControl;
3726
+ /**
3727
+ * Start the music stream.
3728
+ *
3729
+ * @experimental
3730
+ */
3731
+ play(): void;
3732
+ /**
3733
+ * Temporarily halt the music stream. Use `play` to resume from the current
3734
+ * position.
3735
+ *
3736
+ * @experimental
3737
+ */
3738
+ pause(): void;
3739
+ /**
3740
+ * Stop the music stream and reset the state. Retains the current prompts
3741
+ * and config.
3742
+ *
3743
+ * @experimental
3744
+ */
3745
+ stop(): void;
3746
+ /**
3747
+ * Resets the context of the music generation without stopping it.
3748
+ * Retains the current prompts and config.
3749
+ *
3750
+ * @experimental
3751
+ */
3752
+ resetContext(): void;
3753
+ /**
3754
+ Terminates the WebSocket connection.
3755
+
3756
+ @experimental
3757
+ */
3758
+ close(): void;
3759
+ }
3760
+
3761
+ /** Parameters for setting client content for the live API. */
3762
+ export declare interface LiveMusicSetClientContentParameters {
3763
+ /** A map of text prompts to weights to use for the generation request. */
3764
+ weightedPrompts: WeightedPrompt[];
3765
+ }
3766
+
3767
+ /** Parameters for setting config for the live API. */
3768
+ export declare interface LiveMusicSetConfigParameters {
3769
+ /** Configuration for music generation. */
3770
+ musicGenerationConfig: LiveMusicGenerationConfig;
3771
+ }
3772
+
3773
+ /** Prompts and config used for generating this audio chunk. */
3774
+ export declare interface LiveMusicSourceMetadata {
3775
+ /** Weighted prompts for generating this audio chunk. */
3776
+ clientContent?: LiveMusicClientContent;
3777
+ /** Music generation config for generating this audio chunk. */
3778
+ musicGenerationConfig?: LiveMusicGenerationConfig;
3779
+ }
3780
+
3104
3781
  /** Parameters for sending client content to the live API. */
3105
3782
  export declare interface LiveSendClientContentParameters {
3106
3783
  /** Client content to send to the session. */
@@ -3175,6 +3852,8 @@ export declare interface LiveServerContent {
3175
3852
  model turn.
3176
3853
  */
3177
3854
  outputTranscription?: Transcription;
3855
+ /** Metadata related to url context retrieval tool. */
3856
+ urlContextMetadata?: UrlContextMetadata;
3178
3857
  }
3179
3858
 
3180
3859
  /** Server will not be able to service client soon. */
@@ -3324,29 +4003,83 @@ export declare enum MaskReferenceMode {
3324
4003
  MASK_MODE_SEMANTIC = "MASK_MODE_SEMANTIC"
3325
4004
  }
3326
4005
 
4006
+ /**
4007
+ * Creates a McpCallableTool from MCP clients and an optional config.
4008
+ *
4009
+ * The callable tool can invoke the MCP clients with given function call
4010
+ * arguments. (often for automatic function calling).
4011
+ * Use the config to modify tool parameters such as behavior.
4012
+ *
4013
+ * @experimental Built-in MCP support is a preview feature, may change in future
4014
+ * versions.
4015
+ */
4016
+ export declare function mcpToTool(...args: [...Client[], CallableToolConfig | Client]): CallableTool;
4017
+
3327
4018
  /** Server content modalities. */
3328
4019
  export declare enum MediaModality {
4020
+ /**
4021
+ * The modality is unspecified.
4022
+ */
3329
4023
  MODALITY_UNSPECIFIED = "MODALITY_UNSPECIFIED",
4024
+ /**
4025
+ * Plain text.
4026
+ */
3330
4027
  TEXT = "TEXT",
4028
+ /**
4029
+ * Images.
4030
+ */
3331
4031
  IMAGE = "IMAGE",
4032
+ /**
4033
+ * Video.
4034
+ */
3332
4035
  VIDEO = "VIDEO",
4036
+ /**
4037
+ * Audio.
4038
+ */
3333
4039
  AUDIO = "AUDIO",
4040
+ /**
4041
+ * Document, e.g. PDF.
4042
+ */
3334
4043
  DOCUMENT = "DOCUMENT"
3335
4044
  }
3336
4045
 
3337
4046
  /** The media resolution to use. */
3338
4047
  export declare enum MediaResolution {
4048
+ /**
4049
+ * Media resolution has not been set
4050
+ */
3339
4051
  MEDIA_RESOLUTION_UNSPECIFIED = "MEDIA_RESOLUTION_UNSPECIFIED",
4052
+ /**
4053
+ * Media resolution set to low (64 tokens).
4054
+ */
3340
4055
  MEDIA_RESOLUTION_LOW = "MEDIA_RESOLUTION_LOW",
4056
+ /**
4057
+ * Media resolution set to medium (256 tokens).
4058
+ */
3341
4059
  MEDIA_RESOLUTION_MEDIUM = "MEDIA_RESOLUTION_MEDIUM",
4060
+ /**
4061
+ * Media resolution set to high (zoomed reframing with 256 tokens).
4062
+ */
3342
4063
  MEDIA_RESOLUTION_HIGH = "MEDIA_RESOLUTION_HIGH"
3343
4064
  }
3344
4065
 
3345
4066
  /** Server content modalities. */
3346
4067
  export declare enum Modality {
4068
+ /**
4069
+ * The modality is unspecified.
4070
+ */
3347
4071
  MODALITY_UNSPECIFIED = "MODALITY_UNSPECIFIED",
4072
+ /**
4073
+ * Indicates the model should return text
4074
+ */
3348
4075
  TEXT = "TEXT",
4076
+ /**
4077
+ * Indicates the model should return images.
4078
+ */
3349
4079
  IMAGE = "IMAGE",
4080
+ /**
4081
+ * Indicates the model should return images.
4082
+ */
3350
4083
  AUDIO = "AUDIO"
3351
4084
  }
3352
4085
 
@@ -3360,7 +4093,13 @@ export declare interface ModalityTokenCount {
3360
4093
 
3361
4094
  /** The mode of the predictor to be used in dynamic retrieval. */
3362
4095
  export declare enum Mode {
4096
+ /**
4097
+ * Always trigger retrieval.
4098
+ */
3363
4099
  MODE_UNSPECIFIED = "MODE_UNSPECIFIED",
4100
+ /**
4101
+ * Run retrieval only when system decides it is necessary.
4102
+ */
3364
4103
  MODE_DYNAMIC = "MODE_DYNAMIC"
3365
4104
  }
3366
4105
 
@@ -3481,6 +4220,14 @@ export declare class Models extends BaseModule {
3481
4220
  * ```
3482
4221
  */
3483
4222
  generateContentStream: (params: types.GenerateContentParameters) => Promise<AsyncGenerator<types.GenerateContentResponse>>;
4223
+ /**
4224
+ * Transforms the CallableTools in the parameters to be simply Tools, it
4225
+ * copies the params into a new object and replaces the tools, it does not
4226
+ * modify the original params.
4227
+ */
4228
+ private transformCallableTools;
4229
+ private initAfcToolsMap;
4230
+ private processAfcStream;
3484
4231
  /**
3485
4232
  * Generates an image based on a text description and configuration.
3486
4233
  *
@@ -3698,6 +4445,30 @@ export declare interface ModelSelectionConfig {
3698
4445
  featureSelectionPreference?: FeatureSelectionPreference;
3699
4446
  }
3700
4447
 
4448
+ /** The configuration for the multi-speaker setup. */
4449
+ export declare interface MultiSpeakerVoiceConfig {
4450
+ /** The configuration for the speaker to use. */
4451
+ speakerVoiceConfigs?: SpeakerVoiceConfig[];
4452
+ }
4453
+
4454
+ /** The mode of music generation. */
4455
+ export declare enum MusicGenerationMode {
4456
+ /**
4457
+ * This value is unused.
4458
+ */
4459
+ MUSIC_GENERATION_MODE_UNSPECIFIED = "MUSIC_GENERATION_MODE_UNSPECIFIED",
4460
+ /**
4461
+ * Steer text prompts to regions of latent space with higher quality
4462
+ music.
4463
+ */
4464
+ QUALITY = "QUALITY",
4465
+ /**
4466
+ * Steer text prompts to regions of latent space with a larger diversity
4467
+ of music.
4468
+ */
4469
+ DIVERSITY = "DIVERSITY"
4470
+ }
4471
+
3701
4472
  /** A long-running operation. */
3702
4473
  export declare interface Operation {
3703
4474
  /** The server-assigned name, which is only unique within the same service that originally returns it. If you use the default HTTP mapping, the `name` should be a resource name ending with `operations/{unique_id}`. */
@@ -3739,9 +4510,21 @@ export declare class Operations extends BaseModule {
3739
4510
  */
3740
4511
  /** Required. Outcome of the code execution. */
3741
4512
  export declare enum Outcome {
4513
+ /**
4514
+ * Unspecified status. This value should not be used.
4515
+ */
3742
4516
  OUTCOME_UNSPECIFIED = "OUTCOME_UNSPECIFIED",
4517
+ /**
4518
+ * Code execution completed successfully.
4519
+ */
3743
4520
  OUTCOME_OK = "OUTCOME_OK",
4521
+ /**
4522
+ * Code execution finished but with a failure. `stderr` should contain the reason.
4523
+ */
3744
4524
  OUTCOME_FAILED = "OUTCOME_FAILED",
4525
+ /**
4526
+ * Code execution ran for too long, and was cancelled. There may or may not be a partial output present.
4527
+ */
3745
4528
  OUTCOME_DEADLINE_EXCEEDED = "OUTCOME_DEADLINE_EXCEEDED"
3746
4529
  }
3747
4530
 
@@ -3922,11 +4705,18 @@ export declare enum PersonGeneration {
3922
4705
 
3923
4706
  /** The configuration for the prebuilt speaker to use. */
3924
4707
  export declare interface PrebuiltVoiceConfig {
3925
- /** The name of the prebuilt voice to use.
3926
- */
4708
+ /** The name of the prebuilt voice to use. */
3927
4709
  voiceName?: string;
3928
4710
  }
3929
4711
 
4712
+ /** Config for proactivity features. */
4713
+ export declare interface ProactivityConfig {
4714
+ /** If enabled, the model can reject responding to the last prompt. For
4715
+ example, this allows the model to ignore out of context speech or to stay
4716
+ silent if the user did not make a request, yet. */
4717
+ proactiveAudio?: boolean;
4718
+ }
4719
+
3930
4720
  /** Specifies the context retrieval config. */
3931
4721
  export declare interface RagRetrievalConfig {
3932
4722
  /** Optional. Config for filters. */
@@ -4107,6 +4897,62 @@ export declare interface SafetySetting {
4107
4897
  threshold?: HarmBlockThreshold;
4108
4898
  }
4109
4899
 
4900
+ /** Scale of the generated music. */
4901
+ export declare enum Scale {
4902
+ /**
4903
+ * Default value. This value is unused.
4904
+ */
4905
+ SCALE_UNSPECIFIED = "SCALE_UNSPECIFIED",
4906
+ /**
4907
+ * C major or A minor.
4908
+ */
4909
+ C_MAJOR_A_MINOR = "C_MAJOR_A_MINOR",
4910
+ /**
4911
+ * Db major or Bb minor.
4912
+ */
4913
+ D_FLAT_MAJOR_B_FLAT_MINOR = "D_FLAT_MAJOR_B_FLAT_MINOR",
4914
+ /**
4915
+ * D major or B minor.
4916
+ */
4917
+ D_MAJOR_B_MINOR = "D_MAJOR_B_MINOR",
4918
+ /**
4919
+ * Eb major or C minor
4920
+ */
4921
+ E_FLAT_MAJOR_C_MINOR = "E_FLAT_MAJOR_C_MINOR",
4922
+ /**
4923
+ * E major or Db minor.
4924
+ */
4925
+ E_MAJOR_D_FLAT_MINOR = "E_MAJOR_D_FLAT_MINOR",
4926
+ /**
4927
+ * F major or D minor.
4928
+ */
4929
+ F_MAJOR_D_MINOR = "F_MAJOR_D_MINOR",
4930
+ /**
4931
+ * Gb major or Eb minor.
4932
+ */
4933
+ G_FLAT_MAJOR_E_FLAT_MINOR = "G_FLAT_MAJOR_E_FLAT_MINOR",
4934
+ /**
4935
+ * G major or E minor.
4936
+ */
4937
+ G_MAJOR_E_MINOR = "G_MAJOR_E_MINOR",
4938
+ /**
4939
+ * Ab major or F minor.
4940
+ */
4941
+ A_FLAT_MAJOR_F_MINOR = "A_FLAT_MAJOR_F_MINOR",
4942
+ /**
4943
+ * A major or Gb minor.
4944
+ */
4945
+ A_MAJOR_G_FLAT_MINOR = "A_MAJOR_G_FLAT_MINOR",
4946
+ /**
4947
+ * Bb major or G minor.
4948
+ */
4949
+ B_FLAT_MAJOR_G_MINOR = "B_FLAT_MAJOR_G_MINOR",
4950
+ /**
4951
+ * B major or Ab minor.
4952
+ */
4953
+ B_MAJOR_A_FLAT_MINOR = "B_MAJOR_A_FLAT_MINOR"
4954
+ }
4955
+
4110
4956
  /** Schema is used to define the format of input/output data. Represents a select subset of an [OpenAPI 3.0 schema object](https://spec.openapis.org/oas/v3.0.3#schema-object). More fields may be added in the future as needed. */
4111
4957
  export declare interface Schema {
4112
4958
  /** Optional. The value should be validated against any (one or more) of the subschemas in the list. */
@@ -4155,7 +5001,7 @@ export declare interface Schema {
4155
5001
  type?: Type;
4156
5002
  }
4157
5003
 
4158
- export declare type SchemaUnion = Schema;
5004
+ export declare type SchemaUnion = Schema | unknown;
4159
5005
 
4160
5006
  /** Google search entry point. */
4161
5007
  export declare interface SearchEntryPoint {
@@ -4371,11 +5217,24 @@ export declare interface SlidingWindow {
4371
5217
  targetTokens?: string;
4372
5218
  }
4373
5219
 
5220
+ /** The configuration for the speaker to use. */
5221
+ export declare interface SpeakerVoiceConfig {
5222
+ /** The name of the speaker to use. Should be the same as in the
5223
+ prompt. */
5224
+ speaker?: string;
5225
+ /** The configuration for the voice to use. */
5226
+ voiceConfig?: VoiceConfig;
5227
+ }
5228
+
4374
5229
  /** The speech generation configuration. */
4375
5230
  export declare interface SpeechConfig {
4376
5231
  /** The configuration for the speaker to use.
4377
5232
  */
4378
5233
  voiceConfig?: VoiceConfig;
5234
+ /** The configuration for the multi-speaker setup.
5235
+ It is mutually exclusive with the voice_config field.
5236
+ */
5237
+ multiSpeakerVoiceConfig?: MultiSpeakerVoiceConfig;
4379
5238
  /** Language code (ISO 639. e.g. en-US) for the speech synthesization.
4380
5239
  Only available for Live API.
4381
5240
  */
@@ -4386,8 +5245,17 @@ export declare type SpeechConfigUnion = SpeechConfig | string;
4386
5245
 
4387
5246
  /** Start of speech sensitivity. */
4388
5247
  export declare enum StartSensitivity {
5248
+ /**
5249
+ * The default is START_SENSITIVITY_LOW.
5250
+ */
4389
5251
  START_SENSITIVITY_UNSPECIFIED = "START_SENSITIVITY_UNSPECIFIED",
5252
+ /**
5253
+ * Automatic detection will detect the start of speech more often.
5254
+ */
4390
5255
  START_SENSITIVITY_HIGH = "START_SENSITIVITY_HIGH",
5256
+ /**
5257
+ * Automatic detection will detect the start of speech less often.
5258
+ */
4391
5259
  START_SENSITIVITY_LOW = "START_SENSITIVITY_LOW"
4392
5260
  }
4393
5261
 
@@ -4582,6 +5450,8 @@ export declare interface TokensInfo {
4582
5450
 
4583
5451
  /** Tool details of a tool that the model may use to generate a response. */
4584
5452
  export declare interface Tool {
5453
+ /** List of function declarations that the tool supports. */
5454
+ functionDeclarations?: FunctionDeclaration[];
4585
5455
  /** 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. */
4586
5456
  retrieval?: Retrieval;
4587
5457
  /** Optional. Google Search tool type. Specialized retrieval tool
@@ -4595,10 +5465,10 @@ export declare interface Tool {
4595
5465
  /** Optional. Google Maps tool type. Specialized retrieval tool
4596
5466
  that is powered by Google Maps. */
4597
5467
  googleMaps?: GoogleMaps;
5468
+ /** Optional. Tool to support URL context retrieval. */
5469
+ urlContext?: UrlContext;
4598
5470
  /** Optional. CodeExecution tool type. Enables the model to execute code as part of generation. This field is only used by the Gemini Developer API services. */
4599
5471
  codeExecution?: ToolCodeExecution;
4600
- /** 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 128 function declarations can be provided. */
4601
- functionDeclarations?: FunctionDeclaration[];
4602
5472
  }
4603
5473
 
4604
5474
  /** Tool that executes code generated by the model, and automatically returns the result to the model. See also [ExecutableCode]and [CodeExecutionResult] which are input and output to this tool. */
@@ -4616,12 +5486,23 @@ export declare interface ToolConfig {
4616
5486
  retrievalConfig?: RetrievalConfig;
4617
5487
  }
4618
5488
 
4619
- export declare type ToolListUnion = Tool[];
5489
+ export declare type ToolListUnion = ToolUnion[];
5490
+
5491
+ export declare type ToolUnion = Tool | CallableTool;
4620
5492
 
4621
5493
  /** Output only. Traffic type. This shows whether a request consumes Pay-As-You-Go or Provisioned Throughput quota. */
4622
5494
  export declare enum TrafficType {
5495
+ /**
5496
+ * Unspecified request traffic type.
5497
+ */
4623
5498
  TRAFFIC_TYPE_UNSPECIFIED = "TRAFFIC_TYPE_UNSPECIFIED",
5499
+ /**
5500
+ * Type for Pay-As-You-Go traffic.
5501
+ */
4624
5502
  ON_DEMAND = "ON_DEMAND",
5503
+ /**
5504
+ * Type for Provisioned Throughput traffic.
5505
+ */
4625
5506
  PROVISIONED_THROUGHPUT = "PROVISIONED_THROUGHPUT"
4626
5507
  }
4627
5508
 
@@ -4784,19 +5665,49 @@ export declare interface TuningValidationDataset {
4784
5665
 
4785
5666
  /** Options about which input is included in the user's turn. */
4786
5667
  export declare enum TurnCoverage {
5668
+ /**
5669
+ * If unspecified, the default behavior is `TURN_INCLUDES_ONLY_ACTIVITY`.
5670
+ */
4787
5671
  TURN_COVERAGE_UNSPECIFIED = "TURN_COVERAGE_UNSPECIFIED",
5672
+ /**
5673
+ * The users turn only includes activity since the last turn, excluding inactivity (e.g. silence on the audio stream). This is the default behavior.
5674
+ */
4788
5675
  TURN_INCLUDES_ONLY_ACTIVITY = "TURN_INCLUDES_ONLY_ACTIVITY",
5676
+ /**
5677
+ * The users turn includes all realtime input since the last turn, including inactivity (e.g. silence on the audio stream).
5678
+ */
4789
5679
  TURN_INCLUDES_ALL_INPUT = "TURN_INCLUDES_ALL_INPUT"
4790
5680
  }
4791
5681
 
4792
5682
  /** Optional. The type of the data. */
4793
5683
  export declare enum Type {
5684
+ /**
5685
+ * Not specified, should not be used.
5686
+ */
4794
5687
  TYPE_UNSPECIFIED = "TYPE_UNSPECIFIED",
5688
+ /**
5689
+ * OpenAPI string type
5690
+ */
4795
5691
  STRING = "STRING",
5692
+ /**
5693
+ * OpenAPI number type
5694
+ */
4796
5695
  NUMBER = "NUMBER",
5696
+ /**
5697
+ * OpenAPI integer type
5698
+ */
4797
5699
  INTEGER = "INTEGER",
5700
+ /**
5701
+ * OpenAPI boolean type
5702
+ */
4798
5703
  BOOLEAN = "BOOLEAN",
5704
+ /**
5705
+ * OpenAPI array type
5706
+ */
4799
5707
  ARRAY = "ARRAY",
5708
+ /**
5709
+ * OpenAPI object type
5710
+ */
4800
5711
  OBJECT = "OBJECT"
4801
5712
  }
4802
5713
 
@@ -4816,9 +5727,9 @@ declare namespace types {
4816
5727
  HarmCategory,
4817
5728
  HarmBlockMethod,
4818
5729
  HarmBlockThreshold,
5730
+ Type,
4819
5731
  Mode,
4820
5732
  AuthType,
4821
- Type,
4822
5733
  FinishReason,
4823
5734
  HarmProbability,
4824
5735
  HarmSeverity,
@@ -4829,8 +5740,10 @@ declare namespace types {
4829
5740
  JobState,
4830
5741
  AdapterSize,
4831
5742
  FeatureSelectionPreference,
5743
+ Behavior,
4832
5744
  DynamicRetrievalConfigMode,
4833
5745
  FunctionCallingConfigMode,
5746
+ UrlRetrievalStatus,
4834
5747
  SafetyFilterLevel,
4835
5748
  PersonGeneration,
4836
5749
  ImagePromptLanguage,
@@ -4845,8 +5758,12 @@ declare namespace types {
4845
5758
  EndSensitivity,
4846
5759
  ActivityHandling,
4847
5760
  TurnCoverage,
4848
- Blob_2 as Blob,
5761
+ FunctionResponseScheduling,
5762
+ Scale,
5763
+ MusicGenerationMode,
5764
+ LiveMusicPlaybackControl,
4849
5765
  VideoMetadata,
5766
+ Blob_2 as Blob,
4850
5767
  CodeExecutionResult,
4851
5768
  ExecutableCode,
4852
5769
  FileData,
@@ -4857,6 +5774,9 @@ declare namespace types {
4857
5774
  HttpOptions,
4858
5775
  ModelSelectionConfig,
4859
5776
  SafetySetting,
5777
+ Schema,
5778
+ FunctionDeclaration,
5779
+ Interval,
4860
5780
  GoogleSearch,
4861
5781
  DynamicRetrievalConfig,
4862
5782
  GoogleSearchRetrieval,
@@ -4868,6 +5788,7 @@ declare namespace types {
4868
5788
  AuthConfigOidcConfig,
4869
5789
  AuthConfig,
4870
5790
  GoogleMaps,
5791
+ UrlContext,
4871
5792
  VertexAISearch,
4872
5793
  VertexRagStoreRagResource,
4873
5794
  RagRetrievalConfigFilter,
@@ -4879,8 +5800,6 @@ declare namespace types {
4879
5800
  VertexRagStore,
4880
5801
  Retrieval,
4881
5802
  ToolCodeExecution,
4882
- Schema,
4883
- FunctionDeclaration,
4884
5803
  Tool,
4885
5804
  FunctionCallingConfig,
4886
5805
  LatLng,
@@ -4888,7 +5807,10 @@ declare namespace types {
4888
5807
  ToolConfig,
4889
5808
  PrebuiltVoiceConfig,
4890
5809
  VoiceConfig,
5810
+ SpeakerVoiceConfig,
5811
+ MultiSpeakerVoiceConfig,
4891
5812
  SpeechConfig,
5813
+ AutomaticFunctionCallingConfig,
4892
5814
  ThinkingConfig,
4893
5815
  GenerationConfigRoutingConfigAutoRoutingMode,
4894
5816
  GenerationConfigRoutingConfigManualRoutingMode,
@@ -4898,6 +5820,8 @@ declare namespace types {
4898
5820
  GoogleTypeDate,
4899
5821
  Citation,
4900
5822
  CitationMetadata,
5823
+ UrlMetadata,
5824
+ UrlContextMetadata,
4901
5825
  GroundingChunkRetrievedContext,
4902
5826
  GroundingChunkWeb,
4903
5827
  GroundingChunk,
@@ -5016,7 +5940,6 @@ declare namespace types {
5016
5940
  CreateFileParameters,
5017
5941
  HttpResponse,
5018
5942
  LiveCallbacks,
5019
- UploadFileParameters,
5020
5943
  CreateFileResponse,
5021
5944
  GetFileConfig,
5022
5945
  GetFileParameters,
@@ -5058,6 +5981,7 @@ declare namespace types {
5058
5981
  SlidingWindow,
5059
5982
  ContextWindowCompressionConfig,
5060
5983
  AudioTranscriptionConfig,
5984
+ ProactivityConfig,
5061
5985
  LiveClientSetup,
5062
5986
  LiveClientContent,
5063
5987
  ActivityStart,
@@ -5072,6 +5996,24 @@ declare namespace types {
5072
5996
  SendMessageParameters,
5073
5997
  LiveSendClientContentParameters,
5074
5998
  LiveSendToolResponseParameters,
5999
+ LiveMusicClientSetup,
6000
+ WeightedPrompt,
6001
+ LiveMusicClientContent,
6002
+ LiveMusicGenerationConfig,
6003
+ LiveMusicClientMessage,
6004
+ LiveMusicServerSetupComplete,
6005
+ LiveMusicSourceMetadata,
6006
+ AudioChunk,
6007
+ LiveMusicServerContent,
6008
+ LiveMusicFilteredPrompt,
6009
+ LiveMusicServerMessage,
6010
+ LiveMusicCallbacks,
6011
+ UploadFileParameters,
6012
+ CallableTool,
6013
+ CallableToolConfig,
6014
+ LiveMusicConnectParameters,
6015
+ LiveMusicSetConfigParameters,
6016
+ LiveMusicSetClientContentParameters,
5075
6017
  OperationGetParameters,
5076
6018
  BlobImageUnion,
5077
6019
  PartUnion,
@@ -5080,6 +6022,7 @@ declare namespace types {
5080
6022
  ContentListUnion,
5081
6023
  SchemaUnion,
5082
6024
  SpeechConfigUnion,
6025
+ ToolUnion,
5083
6026
  ToolListUnion,
5084
6027
  DownloadableFileUnion
5085
6028
  }
@@ -5226,6 +6169,40 @@ export declare class UpscaleImageResponse {
5226
6169
  generatedImages?: GeneratedImage[];
5227
6170
  }
5228
6171
 
6172
+ /** Tool to support URL context retrieval. */
6173
+ export declare interface UrlContext {
6174
+ }
6175
+
6176
+ /** Metadata related to url context retrieval tool. */
6177
+ export declare interface UrlContextMetadata {
6178
+ /** List of url context. */
6179
+ urlMetadata?: UrlMetadata[];
6180
+ }
6181
+
6182
+ /** Context for a single url retrieval. */
6183
+ export declare interface UrlMetadata {
6184
+ /** The URL retrieved by the tool. */
6185
+ retrievedUrl?: string;
6186
+ /** Status of the url retrieval. */
6187
+ urlRetrievalStatus?: UrlRetrievalStatus;
6188
+ }
6189
+
6190
+ /** Status of the url retrieval. */
6191
+ export declare enum UrlRetrievalStatus {
6192
+ /**
6193
+ * Default value. This value is unused
6194
+ */
6195
+ URL_RETRIEVAL_STATUS_UNSPECIFIED = "URL_RETRIEVAL_STATUS_UNSPECIFIED",
6196
+ /**
6197
+ * Url retrieval is successful.
6198
+ */
6199
+ URL_RETRIEVAL_STATUS_SUCCESS = "URL_RETRIEVAL_STATUS_SUCCESS",
6200
+ /**
6201
+ * Url retrieval is failed due to error.
6202
+ */
6203
+ URL_RETRIEVAL_STATUS_ERROR = "URL_RETRIEVAL_STATUS_ERROR"
6204
+ }
6205
+
5229
6206
  /** Usage metadata about response(s). */
5230
6207
  export declare interface UsageMetadata {
5231
6208
  /** Number of tokens in the prompt. When `cached_content` is set, this is still the total effective prompt size meaning this includes the number of tokens in the cached content. */
@@ -5293,8 +6270,11 @@ export declare interface Video {
5293
6270
  mimeType?: string;
5294
6271
  }
5295
6272
 
5296
- /** Metadata describes the input video content. */
6273
+ /** Describes how the video in the Part should be used by the model. */
5297
6274
  export declare interface VideoMetadata {
6275
+ /** The frame rate of the video sent to the model. If not specified, the
6276
+ default value will be 1.0. The fps range is (0.0, 24.0]. */
6277
+ fps?: number;
5298
6278
  /** Optional. The end offset of the video. */
5299
6279
  endOffset?: string;
5300
6280
  /** Optional. The start offset of the video. */
@@ -5342,4 +6322,17 @@ declare interface WebSocketFactory {
5342
6322
  create(url: string, headers: Record<string, string>, callbacks: WebSocketCallbacks): WebSocket_2;
5343
6323
  }
5344
6324
 
6325
+ /** Maps a prompt to a relative weight to steer music generation. */
6326
+ export declare interface WeightedPrompt {
6327
+ /** Text prompt. */
6328
+ text?: string;
6329
+ /** Weight of the prompt. The weight is used to control the relative
6330
+ importance of the prompt. Higher weights are more important than lower
6331
+ weights.
6332
+
6333
+ Weight must not be 0. Weights of all weighted_prompts in this
6334
+ LiveMusicClientContent message will be normalized. */
6335
+ weight?: number;
6336
+ }
6337
+
5345
6338
  export { }