@google/genai 2.12.0 → 2.13.0

package/dist/web/web.d.ts

@@ -80,6 +80,10 @@ declare type Agent$ = Agent;
  *   }
  */
 declare type Agent = {
+    /**
+     * Configuration parameters for the agent.
+     */
+    agent_config?: interactions.AntigravityAgentConfig | undefined;
     /**
      * The base agent to extend.
      */
@@ -106,6 +110,11 @@ declare type Agent = {
     tools?: Array<AgentTool> | undefined;
 };
 
+/**
+ * Configuration parameters for the agent.
+ */
+declare type AgentConfig = interactions.AntigravityAgentConfig;
+
 declare type AgentCreateParams$ = CreateAgentParams;
 
 declare type AgentDeleteParams$ = DeleteAgentParams;
@@ -150,6 +159,7 @@ declare namespace agents {
     export {
         AgentListResponse,
         AgentTool,
+        AgentConfig,
         BaseEnvironment,
         Agent
     }
@@ -284,7 +294,7 @@ declare type Annotation$ = Annotation;
 /**
  * Citation information for model-generated content.
  */
-declare type Annotation = URLCitation | FileCitation | PlaceCitation;
+declare type Annotation = URLCitation | FileCitation | PlaceCitation | WordInfo;
 
 declare type AntigravityAgentConfig$ = AntigravityAgentConfig;
 
@@ -307,6 +317,10 @@ declare type AntigravityAgentConfig = {
      * Max total tokens for the agent run.
      */
     max_total_tokens?: string | undefined;
+    /**
+     * The model to use for agent reasoning.
+     */
+    model?: string | undefined;
     type: "antigravity";
 };
 
@@ -744,7 +758,9 @@ export declare interface AudioTranscriptionConfig {
     languageAuto?: LanguageAuto;
     /** Specifies one or more languages in the audio. Do not use together with LanguageAuto. */
     languageHints?: LanguageHints;
-    /** A list of phrases used for speech adaptation, which biases the ASR model to improve recognition of these specific terms. */
+    /** A list of custom vocabulary phrases, which biases the ASR model to improve recognition of these specific terms. */
+    customVocabulary?: string[];
+    /** Deprecated. A list of phrases used for speech adaptation, which biases the ASR model to improve recognition of these specific terms. */
     adaptationPhrases?: string[];
 }
 
@@ -4815,7 +4831,7 @@ declare type FunctionResultStep = {
      */
     name?: string | undefined;
     /**
-     * The result of the tool call.
+     * Required. The result of the tool call.
      */
     result: FunctionResultStepResult | Array<FunctionResultSubcontent> | string;
     type: "function_result";
@@ -4824,7 +4840,7 @@ declare type FunctionResultStep = {
 declare type FunctionResultStepResult = {};
 
 /**
- * The result of the tool call.
+ * Required. The result of the tool call.
  */
 declare type FunctionResultStepResultUnion = FunctionResultStepResult | Array<FunctionResultSubcontent> | string;
 
@@ -5644,6 +5660,10 @@ declare type GenerationConfig_2 = {
      * The maximum cumulative probability of tokens to consider when sampling.
      */
     top_p?: number | undefined;
+    /**
+     * Configuration for speech recognition (transcription).
+     */
+    transcription_config?: TranscriptionConfig | undefined;
     /**
      * Configuration options for video generation.
      */
@@ -7611,6 +7631,7 @@ export declare namespace Interactions {
     export type Tool = Tool$;
     export type ToolChoiceConfig = ToolChoiceConfig$;
     export type ToolChoiceType = ToolChoiceType$;
+    export type TranscriptionConfig = TranscriptionConfig$;
     export type URLCitation = URLCitation$;
     export type URLContextCallArguments = URLContextCallArguments$;
     export interface URLContextCallStep extends URLContextCallStep$ {
@@ -7625,6 +7646,7 @@ export declare namespace Interactions {
     export type VideoContent = VideoContent$;
     export type VideoResponseFormat = VideoResponseFormat$;
     export type WebhookConfig = WebhookConfig$;
+    export type WordInfo = WordInfo$;
     export namespace CodeExecutionCallStep {
         export type Arguments = Arguments$;
     }
@@ -7938,6 +7960,7 @@ declare namespace interactions {
         ToolChoiceConfig,
         ToolChoiceType,
         Tool_2 as Tool,
+        TranscriptionConfig,
         TurnContent,
         Turn,
         URLCitation,
@@ -7961,7 +7984,8 @@ declare namespace interactions {
         VideoResponseFormatAspectRatio,
         VideoResponseFormatDelivery,
         VideoResponseFormat_2 as VideoResponseFormat,
-        WebhookConfig_2 as WebhookConfig
+        WebhookConfig_2 as WebhookConfig,
+        WordInfo
     }
 }
 
@@ -8033,7 +8057,7 @@ declare type InteractionSSEStreamEvent = {
 /**
  * Required. Output only. The status of the interaction.
  */
-declare type InteractionStatus = "in_progress" | "requires_action" | "completed" | "failed" | "cancelled" | "incomplete" | "budget_exceeded" | (string & {});
+declare type InteractionStatus = "in_progress" | "requires_action" | "completed" | "failed" | "cancelled" | "incomplete" | "budget_exceeded" | "queued" | (string & {});
 
 declare type InteractionStatusUpdate$ = InteractionStatusUpdate;
 
@@ -8051,7 +8075,7 @@ declare type InteractionStatusUpdate = {
     status: InteractionStatusUpdateStatus;
 };
 
-declare type InteractionStatusUpdateStatus = "in_progress" | "requires_action" | "completed" | "failed" | "cancelled" | "incomplete" | "budget_exceeded" | (string & {});
+declare type InteractionStatusUpdateStatus = "in_progress" | "requires_action" | "completed" | "failed" | "cancelled" | "incomplete" | "budget_exceeded" | "queued" | (string & {});
 
 /** Parameters for the private _Register method. */
 export declare interface InternalRegisterFilesParameters {
@@ -9523,7 +9547,7 @@ declare type MCPServerToolResultStep = {
      */
     name?: string | undefined;
     /**
-     * The output from the MCP server call. Can be simple text or rich content.
+     * Required. The output from the MCP server call. Can be simple text or rich content.
      */
     result: MCPServerToolResultStepResult | string | Array<FunctionResultSubcontent>;
     /**
@@ -9536,7 +9560,7 @@ declare type MCPServerToolResultStep = {
 declare type MCPServerToolResultStepResult = {};
 
 /**
- * The output from the MCP server call. Can be simple text or rich content.
+ * Required. The output from the MCP server call. Can be simple text or rich content.
  */
 declare type MCPServerToolResultStepResultUnion = MCPServerToolResultStepResult | string | Array<FunctionResultSubcontent>;
 
@@ -13264,6 +13288,53 @@ export declare interface Transcription {
     languageCode?: string;
 }
 
+declare type TranscriptionConfig$ = TranscriptionConfig;
+
+/**
+ * @license
+ * Copyright 2026 Google LLC
+ * SPDX-License-Identifier: Apache-2.0
+ *
+ * g3-prettier-ignore-file
+ */
+/**
+ * Configuration for speech recognition (transcription).
+ */
+declare type TranscriptionConfig = {
+    /**
+     * Optional. A list of phrases to bias the ASR model towards.
+     *
+     * @deprecated field: This will be removed in a future release, please migrate away from it as soon as possible.
+     */
+    adaptation_phrases?: Array<string> | undefined;
+    /**
+     * Optional. A list of custom vocabulary phrases to bias the speech recognition model
+     *
+     * @remarks
+     * toward recognizing specific terms.
+     */
+    custom_vocabulary?: Array<string> | undefined;
+    /**
+     * Optional. Configures speaker diarization. Supported values: "speaker".
+     */
+    diarization_mode?: string | undefined;
+    /**
+     * Required. BCP-47 language codes providing hints about the languages present in the
+     *
+     * @remarks
+     * audio. At least one must be specified, or set to ["auto"] to enable
+     * automatic language detection.
+     */
+    language_hints: Array<string>;
+    /**
+     * Optional. The granularity of timestamps to include in the transcription output.
+     *
+     * @remarks
+     * Supported values: "word". If empty, no timestamps are generated.
+     */
+    timestamp_granularities?: Array<string> | undefined;
+};
+
 /**
  * @license
  * Copyright 2026 Google LLC
@@ -16008,4 +16079,60 @@ export declare interface WhiteSpaceConfig {
     maxOverlapTokens?: number;
 }
 
+declare type WordInfo$ = WordInfo;
+
+/**
+ * @license
+ * Copyright 2026 Google LLC
+ * SPDX-License-Identifier: Apache-2.0
+ *
+ * g3-prettier-ignore-file
+ */
+/**
+ * Word-level ASR annotation for transcription output.
+ *
+ * @remarks
+ * Carries the word text, optional timing, and optional speaker attribution.
+ */
+declare type WordInfo = {
+    /**
+     * End of the attributed segment, exclusive.
+     */
+    end_index?: number | undefined;
+    /**
+     * End offset in time of the word relative to the start of the audio.
+     *
+     * @remarks
+     * Present when timestamp_granularities contains "word".
+     */
+    end_offset?: string | undefined;
+    /**
+     * Optional. Speaker label for this word (e.g. "spk_1", "spk_2").
+     *
+     * @remarks
+     * Present when diarization_mode is set in TranscriptionConfig.
+     */
+    speaker?: string | undefined;
+    /**
+     * Start of segment of the response that is attributed to this source.
+     *
+     * @remarks
+     *
+     * Index indicates the start of the segment, measured in bytes.
+     */
+    start_index?: number | undefined;
+    /**
+     * Start offset in time of the word relative to the start of the audio.
+     *
+     * @remarks
+     * Present when timestamp_granularities contains "word".
+     */
+    start_offset?: string | undefined;
+    /**
+     * The transcribed word.
+     */
+    text?: string | undefined;
+    type: "word_info";
+};
+
 export { }

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@google/genai",
-  "version": "2.12.0",
+  "version": "2.13.0",
   "description": "",
   "type": "module",
   "main": "dist/node/index.mjs",