@google/genai 0.8.0 → 0.9.0

package/dist/web/web.d.ts

@@ -46,11 +46,11 @@ declare class ApiClient {
     private shouldPrependVertexProjectPath;
     request(request: HttpRequest): Promise<HttpResponse>;
     private patchHttpOptions;
-    requestStream(request: HttpRequest): Promise<any>;
+    requestStream(request: HttpRequest): Promise<AsyncGenerator<HttpResponse>>;
     private includeExtraHttpOptionsToRequestInit;
     private unaryApiCall;
     private streamApiCall;
-    processStreamResponse(response: Response): AsyncGenerator<any>;
+    processStreamResponse(response: Response): AsyncGenerator<HttpResponse>;
     private apiCall;
     getDefaultHeaders(): Record<string, string>;
     private getHeadersInternal;
@@ -237,8 +237,8 @@ export declare class Caches extends BaseModule {
      *
      * @remarks
      * Context caching is only supported for specific models. See [Gemini
-     * Developer API reference] (https://ai.google.dev/gemini-api/docs/caching?lang=node/context-cac)
-     * and [Vertex AI reference] (https://cloud.google.com/vertex-ai/generative-ai/docs/context-cache/context-cache-overview#supported_models)
+     * Developer API reference](https://ai.google.dev/gemini-api/docs/caching?lang=node/context-cac)
+     * and [Vertex AI reference](https://cloud.google.com/vertex-ai/generative-ai/docs/context-cache/context-cache-overview#supported_models)
      * for more information.
      *
      * @param params - The parameters for the create request.
@@ -540,7 +540,7 @@ export declare interface ContentEmbeddingStatistics {
     tokenCount?: number;
 }
 
-export declare type ContentListUnion = ContentUnion[] | ContentUnion;
+export declare type ContentListUnion = Content | Content[] | PartUnion | PartUnion[];
 
 export declare type ContentUnion = Content | PartUnion[] | PartUnion;
 
@@ -897,6 +897,14 @@ export declare interface ExecutableCode {
     language?: Language;
 }
 
+/** Options for feature selection preference. */
+export declare enum FeatureSelectionPreference {
+    FEATURE_SELECTION_PREFERENCE_UNSPECIFIED = "FEATURE_SELECTION_PREFERENCE_UNSPECIFIED",
+    PRIORITIZE_QUALITY = "PRIORITIZE_QUALITY",
+    BALANCED = "BALANCED",
+    PRIORITIZE_COST = "PRIORITIZE_COST"
+}
+
 export declare interface FetchPredictOperationConfig {
     /** Used to override HTTP request options. */
     httpOptions?: HttpOptions;
@@ -1239,6 +1247,9 @@ export declare interface GenerateContentConfig {
     /** Configuration for model router requests.
      */
     routingConfig?: GenerationConfigRoutingConfig;
+    /** Configuration for model selection.
+     */
+    modelSelectionConfig?: ModelSelectionConfig;
     /** Safety settings in the request to block unsafe content in the
      response.
      */
@@ -2133,17 +2144,20 @@ export declare class Live {
      @experimental
 
      @remarks
-     If using the Gemini API, Live is currently only supported behind API
-     version `v1alpha`. Ensure that the API version is set to `v1alpha` when
-     initializing the SDK if relying on the Gemini API.
 
      @param params - The parameters for establishing a connection to the model.
      @return A live session.
 
      @example
      ```ts
+     let model: string;
+     if (GOOGLE_GENAI_USE_VERTEXAI) {
+     model = 'gemini-2.0-flash-live-preview-04-09';
+     } else {
+     model = 'gemini-2.0-flash-live-001';
+     }
      const session = await ai.live.connect({
-     model: 'gemini-2.0-flash-exp',
+     model: model,
      config: {
      responseModalities: [Modality.AUDIO],
      },
@@ -2258,7 +2272,7 @@ export declare interface LiveClientSetup {
     /** The user provided system instructions for the model.
      Note: only text should be used in parts and content in each part will be
      in a separate paragraph. */
-    systemInstruction?: Content;
+    systemInstruction?: ContentUnion;
     /**  A list of `Tools` the model may use to generate the next response.
 
      A `Tool` is a piece of code that enables the system to interact with
@@ -2275,6 +2289,13 @@ export declare interface LiveClientSetup {
 
      If included, server will compress context window to fit into given length. */
     contextWindowCompression?: ContextWindowCompressionConfig;
+    /** The transcription of the input aligns with the input audio language.
+     */
+    inputAudioTranscription?: AudioTranscriptionConfig;
+    /** The transcription of the output aligns with the language code
+     specified for the output audio.
+     */
+    outputAudioTranscription?: AudioTranscriptionConfig;
 }
 
 /** Client generated response to a `ToolCall` received from the server.
@@ -2300,13 +2321,42 @@ export declare interface LiveConnectConfig {
      modalities that the model can return. Defaults to AUDIO if not specified.
      */
     responseModalities?: Modality[];
+    /** Value that controls the degree of randomness in token selection.
+     Lower temperatures are good for prompts that require a less open-ended or
+     creative response, while higher temperatures can lead to more diverse or
+     creative results.
+     */
+    temperature?: number;
+    /** Tokens are selected from the most to least probable until the sum
+     of their probabilities equals this value. Use a lower value for less
+     random responses and a higher value for more random responses.
+     */
+    topP?: number;
+    /** For each token selection step, the ``top_k`` tokens with the
+     highest probabilities are sampled. Then tokens are further filtered based
+     on ``top_p`` with the final token selected using temperature sampling. Use
+     a lower number for less random responses and a higher number for more
+     random responses.
+     */
+    topK?: number;
+    /** Maximum number of tokens that can be generated in the response.
+     */
+    maxOutputTokens?: number;
+    /** If specified, the media resolution specified will be used.
+     */
+    mediaResolution?: MediaResolution;
+    /** When ``seed`` is fixed to a specific number, the model makes a best
+     effort to provide the same response for repeated requests. By default, a
+     random number is used.
+     */
+    seed?: number;
     /** The speech generation configuration.
      */
     speechConfig?: SpeechConfig;
     /** The user provided system instructions for the model.
      Note: only text should be used in parts and content in each part will be
      in a separate paragraph. */
-    systemInstruction?: Content;
+    systemInstruction?: ContentUnion;
     /** A list of `Tools` the model may use to generate the next response.
 
      A `Tool` is a piece of code that enables the system to interact with
@@ -2317,6 +2367,13 @@ export declare interface LiveConnectConfig {
 
      If included the server will send SessionResumptionUpdate messages. */
     sessionResumption?: SessionResumptionConfig;
+    /** The transcription of the input aligns with the input audio language.
+     */
+    inputAudioTranscription?: AudioTranscriptionConfig;
+    /** The transcription of the output aligns with the language code
+     specified for the output audio.
+     */
+    outputAudioTranscription?: AudioTranscriptionConfig;
     /** Configures the realtime input behavior in BidiGenerateContent. */
     realtimeInputConfig?: RealtimeInputConfig;
     /** Configures context window compression mechanism.
@@ -2384,6 +2441,15 @@ export declare interface LiveServerContent {
      has finished generating all content. This is a signal to the client
      that it can stop sending messages. */
     generationComplete?: boolean;
+    /** Input transcription. The transcription is independent to the model
+     turn which means it doesn’t imply any ordering between transcription and
+     model turn. */
+    inputTranscription?: Transcription;
+    /** Output transcription. The transcription is independent to the model
+     turn which means it doesn’t imply any ordering between transcription and
+     model turn.
+     */
+    outputTranscription?: Transcription;
 }
 
 /** Server will not be able to service client soon. */
@@ -2804,6 +2870,12 @@ export declare class Models extends BaseModule {
     generateVideos(params: types.GenerateVideosParameters): Promise<types.GenerateVideosOperation>;
 }
 
+/** Config for model selection. */
+export declare interface ModelSelectionConfig {
+    /** Options for feature selection preference. */
+    featureSelectionPreference?: FeatureSelectionPreference;
+}
+
 /** Parameters for the get method of the operations module. */
 export declare interface OperationGetParameters {
     /** The operation to be retrieved. */
@@ -3011,6 +3083,54 @@ export declare interface PrebuiltVoiceConfig {
     voiceName?: string;
 }
 
+/** Specifies the context retrieval config. */
+export declare interface RagRetrievalConfig {
+    /** Optional. Config for filters. */
+    filter?: RagRetrievalConfigFilter;
+    /** Optional. Config for Hybrid Search. */
+    hybridSearch?: RagRetrievalConfigHybridSearch;
+    /** Optional. Config for ranking and reranking. */
+    ranking?: RagRetrievalConfigRanking;
+    /** Optional. The number of contexts to retrieve. */
+    topK?: number;
+}
+
+/** Config for filters. */
+export declare interface RagRetrievalConfigFilter {
+    /** Optional. String for metadata filtering. */
+    metadataFilter?: string;
+    /** Optional. Only returns contexts with vector distance smaller than the threshold. */
+    vectorDistanceThreshold?: number;
+    /** Optional. Only returns contexts with vector similarity larger than the threshold. */
+    vectorSimilarityThreshold?: number;
+}
+
+/** Config for Hybrid Search. */
+export declare interface RagRetrievalConfigHybridSearch {
+    /** Optional. Alpha value controls the weight between dense and sparse vector search results. The range is [0, 1], while 0 means sparse vector search only and 1 means dense vector search only. The default value is 0.5 which balances sparse and dense vector search equally. */
+    alpha?: number;
+}
+
+/** Config for ranking and reranking. */
+export declare interface RagRetrievalConfigRanking {
+    /** Optional. Config for LlmRanker. */
+    llmRanker?: RagRetrievalConfigRankingLlmRanker;
+    /** Optional. Config for Rank Service. */
+    rankService?: RagRetrievalConfigRankingRankService;
+}
+
+/** Config for LlmRanker. */
+export declare interface RagRetrievalConfigRankingLlmRanker {
+    /** Optional. The model name used for ranking. Format: `gemini-1.5-pro` */
+    modelName?: string;
+}
+
+/** Config for Rank Service. */
+export declare interface RagRetrievalConfigRankingRankService {
+    /** Optional. The model name of the rank service. Format: `semantic-ranker-512@latest` */
+    modelName?: string;
+}
+
 /** A raw reference image.
 
  A raw reference image represents the base image to edit, provided by the user.
@@ -3333,8 +3453,14 @@ export declare class Session {
 
      @example
      ```ts
+     let model: string;
+     if (GOOGLE_GENAI_USE_VERTEXAI) {
+     model = 'gemini-2.0-flash-live-preview-04-09';
+     } else {
+     model = 'gemini-2.0-flash-live-001';
+     }
      const session = await ai.live.connect({
-     model: 'gemini-2.0-flash-exp',
+     model: model,
      config: {
      responseModalities: [Modality.AUDIO],
      }
@@ -3377,6 +3503,10 @@ export declare interface SpeechConfig {
     /** The configuration for the speaker to use.
      */
     voiceConfig?: VoiceConfig;
+    /** Language code (ISO 639. e.g. en-US) for the speech synthesization.
+     Only available for Live API.
+     */
+    languageCode?: string;
 }
 
 export declare type SpeechConfigUnion = SpeechConfig | string;
@@ -3531,6 +3661,16 @@ export declare enum TrafficType {
     PROVISIONED_THROUGHPUT = "PROVISIONED_THROUGHPUT"
 }
 
+/** Audio transcription in Server Conent. */
+export declare interface Transcription {
+    /** Transcription text.
+     */
+    text?: string;
+    /** The bool indicates the end of the transcription.
+     */
+    finished?: boolean;
+}
+
 /** A tuned machine learning model. */
 export declare interface TunedModelInfo {
     /** ID of the base model that you want to tune. */
@@ -3584,6 +3724,7 @@ declare namespace types {
         TrafficType,
         Modality,
         MediaResolution,
+        FeatureSelectionPreference,
         DynamicRetrievalConfigMode,
         FunctionCallingConfigMode,
         SafetyFilterLevel,
@@ -3610,6 +3751,7 @@ declare namespace types {
         Content,
         HttpOptions,
         Schema,
+        ModelSelectionConfig,
         SafetySetting,
         FunctionDeclaration,
         GoogleSearch,
@@ -3617,6 +3759,12 @@ declare namespace types {
         GoogleSearchRetrieval,
         VertexAISearch,
         VertexRagStoreRagResource,
+        RagRetrievalConfigFilter,
+        RagRetrievalConfigHybridSearch,
+        RagRetrievalConfigRankingLlmRanker,
+        RagRetrievalConfigRankingRankService,
+        RagRetrievalConfigRanking,
+        RagRetrievalConfig,
         VertexRagStore,
         Retrieval,
         ToolCodeExecution,
@@ -3737,6 +3885,7 @@ declare namespace types {
         SubjectReferenceConfig,
         SubjectReferenceImage,
         LiveServerSetupComplete,
+        Transcription,
         LiveServerContent,
         LiveServerToolCall,
         LiveServerToolCallCancellation,
@@ -3749,6 +3898,7 @@ declare namespace types {
         SessionResumptionConfig,
         SlidingWindow,
         ContextWindowCompressionConfig,
+        AudioTranscriptionConfig,
         LiveClientSetup,
         LiveClientContent,
         ActivityStart,
@@ -3756,7 +3906,6 @@ declare namespace types {
         LiveClientRealtimeInput,
         LiveClientToolResponse,
         LiveClientMessage,
-        AudioTranscriptionConfig,
         LiveConnectConfig,
         LiveConnectParameters,
         CreateChatParameters,
@@ -3909,6 +4058,8 @@ export declare interface VertexRagStore {
     ragCorpora?: string[];
     /** Optional. The representation of the rag source. It can be used to specify corpus only or ragfiles. Currently only support one corpus or multiple files from one corpus. In the future we may open up multiple corpora support. */
     ragResources?: VertexRagStoreRagResource[];
+    /** Optional. The retrieval config for the Rag query. */
+    ragRetrievalConfig?: RagRetrievalConfig;
     /** Optional. Number of top k results to return from the selected corpora. */
     similarityTopK?: number;
     /** Optional. Only return results with vector distance smaller than the threshold. */

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@google/genai",
-  "version": "0.8.0",
+  "version": "0.9.0",
   "description": "",
   "main": "dist/node/index.js",
   "module": "dist/web/index.mjs",
@@ -93,7 +93,9 @@
   },
   "dependencies": {
     "google-auth-library": "^9.14.2",
-    "ws": "^8.18.0"
+    "ws": "^8.18.0",
+    "zod": "^3.22.4",
+    "zod-to-json-schema": "^3.22.4"
   },
   "repository": {
     "type": "git",