@ai-sdk/google 3.0.118 → 3.0.120

package/CHANGELOG.md

@@ -1,5 +1,19 @@
 # @ai-sdk/google
 
+## 3.0.120
+
+### Patch Changes
+
+- daaa769: fix(google): support video response formats in Gemini Interactions
+
+## 3.0.119
+
+### Patch Changes
+
+- 1912e88: Preserve complete Google Generative Language usage metadata in raw usage results.
+- Updated dependencies [cc23556]
+  - @ai-sdk/provider-utils@4.0.50
+
 ## 3.0.118
 
 ### Patch Changes

package/dist/index.d.mts

@@ -1,7 +1,7 @@
 import * as _ai_sdk_provider_utils from '@ai-sdk/provider-utils';
 import { InferSchema, Resolvable, FetchFunction } from '@ai-sdk/provider-utils';
-import { TranscriptionModelV3, ProviderV3, LanguageModelV3, ImageModelV3, EmbeddingModelV3, Experimental_VideoModelV3 } from '@ai-sdk/provider';
 import { z } from 'zod/v4';
+import { TranscriptionModelV3, ProviderV3, LanguageModelV3, ImageModelV3, EmbeddingModelV3, Experimental_VideoModelV3 } from '@ai-sdk/provider';
 
 declare const googleErrorDataSchema: _ai_sdk_provider_utils.LazySchema<{
     error: {
@@ -174,18 +174,32 @@ declare const responseSchema: _ai_sdk_provider_utils.LazySchema<{
         } | null | undefined;
     }[] | null | undefined;
     usageMetadata?: {
+        [x: string]: z.core.util.JSONType;
         cachedContentTokenCount?: number | null | undefined;
         thoughtsTokenCount?: number | null | undefined;
         promptTokenCount?: number | null | undefined;
         candidatesTokenCount?: number | null | undefined;
+        toolUsePromptTokenCount?: number | null | undefined;
         totalTokenCount?: number | null | undefined;
         trafficType?: string | null | undefined;
         serviceTier?: string | null | undefined;
         promptTokensDetails?: {
+            [x: string]: z.core.util.JSONType;
+            modality: string;
+            tokenCount: number;
+        }[] | null | undefined;
+        cacheTokensDetails?: {
+            [x: string]: z.core.util.JSONType;
             modality: string;
             tokenCount: number;
         }[] | null | undefined;
         candidatesTokensDetails?: {
+            [x: string]: z.core.util.JSONType;
+            modality: string;
+            tokenCount: number;
+        }[] | null | undefined;
+        toolUsePromptTokensDetails?: {
+            [x: string]: z.core.util.JSONType;
             modality: string;
             tokenCount: number;
         }[] | null | undefined;
@@ -330,6 +344,14 @@ declare const googleInteractionsLanguageModelOptions: _ai_sdk_provider_utils.Laz
         [x: string]: unknown;
         type: "audio";
         mimeType?: string | null | undefined;
+    } | {
+        [x: string]: unknown;
+        type: "video";
+        aspectRatio?: "9:16" | "16:9" | null | undefined;
+        resolution?: "720p" | "1080p" | "4k" | "360p" | null | undefined;
+        duration?: string | null | undefined;
+        delivery?: "uri" | "inline" | null | undefined;
+        gcsUri?: string | null | undefined;
     })[] | null | undefined;
     imageConfig?: {
         aspectRatio?: "1:1" | "2:3" | "3:2" | "3:4" | "4:3" | "4:5" | "5:4" | "9:16" | "16:9" | "21:9" | "1:8" | "8:1" | "1:4" | "4:1" | null | undefined;

package/dist/index.d.ts

@@ -1,7 +1,7 @@
 import * as _ai_sdk_provider_utils from '@ai-sdk/provider-utils';
 import { InferSchema, Resolvable, FetchFunction } from '@ai-sdk/provider-utils';
-import { TranscriptionModelV3, ProviderV3, LanguageModelV3, ImageModelV3, EmbeddingModelV3, Experimental_VideoModelV3 } from '@ai-sdk/provider';
 import { z } from 'zod/v4';
+import { TranscriptionModelV3, ProviderV3, LanguageModelV3, ImageModelV3, EmbeddingModelV3, Experimental_VideoModelV3 } from '@ai-sdk/provider';
 
 declare const googleErrorDataSchema: _ai_sdk_provider_utils.LazySchema<{
     error: {
@@ -174,18 +174,32 @@ declare const responseSchema: _ai_sdk_provider_utils.LazySchema<{
         } | null | undefined;
     }[] | null | undefined;
     usageMetadata?: {
+        [x: string]: z.core.util.JSONType;
         cachedContentTokenCount?: number | null | undefined;
         thoughtsTokenCount?: number | null | undefined;
         promptTokenCount?: number | null | undefined;
         candidatesTokenCount?: number | null | undefined;
+        toolUsePromptTokenCount?: number | null | undefined;
         totalTokenCount?: number | null | undefined;
         trafficType?: string | null | undefined;
         serviceTier?: string | null | undefined;
         promptTokensDetails?: {
+            [x: string]: z.core.util.JSONType;
+            modality: string;
+            tokenCount: number;
+        }[] | null | undefined;
+        cacheTokensDetails?: {
+            [x: string]: z.core.util.JSONType;
             modality: string;
             tokenCount: number;
         }[] | null | undefined;
         candidatesTokensDetails?: {
+            [x: string]: z.core.util.JSONType;
+            modality: string;
+            tokenCount: number;
+        }[] | null | undefined;
+        toolUsePromptTokensDetails?: {
+            [x: string]: z.core.util.JSONType;
             modality: string;
             tokenCount: number;
         }[] | null | undefined;
@@ -330,6 +344,14 @@ declare const googleInteractionsLanguageModelOptions: _ai_sdk_provider_utils.Laz
         [x: string]: unknown;
         type: "audio";
         mimeType?: string | null | undefined;
+    } | {
+        [x: string]: unknown;
+        type: "video";
+        aspectRatio?: "9:16" | "16:9" | null | undefined;
+        resolution?: "720p" | "1080p" | "4k" | "360p" | null | undefined;
+        duration?: string | null | undefined;
+        delivery?: "uri" | "inline" | null | undefined;
+        gcsUri?: string | null | undefined;
     })[] | null | undefined;
     imageConfig?: {
         aspectRatio?: "1:1" | "2:3" | "3:2" | "3:4" | "4:3" | "4:5" | "5:4" | "9:16" | "16:9" | "21:9" | "1:8" | "8:1" | "1:4" | "4:1" | null | undefined;

package/dist/index.js

@@ -235,7 +235,7 @@ var googleInteractionsTranscriptionResponseSchema = import_v43.z.object({
 var import_provider_utils24 = require("@ai-sdk/provider-utils");
 
 // src/version.ts
-var VERSION = true ? "3.0.118" : "0.0.0-test";
+var VERSION = true ? "3.0.120" : "0.0.0-test";
 
 // src/google-generative-ai-embedding-model.ts
 var import_provider2 = require("@ai-sdk/provider");
@@ -2907,21 +2907,24 @@ var tokenDetailsSchema = import_v47.z.array(
   import_v47.z.object({
     modality: import_v47.z.string(),
     tokenCount: import_v47.z.number()
-  })
+  }).catchall(import_v47.z.json())
 ).nullish();
 var usageSchema = import_v47.z.object({
   cachedContentTokenCount: import_v47.z.number().nullish(),
   thoughtsTokenCount: import_v47.z.number().nullish(),
   promptTokenCount: import_v47.z.number().nullish(),
   candidatesTokenCount: import_v47.z.number().nullish(),
+  toolUsePromptTokenCount: import_v47.z.number().nullish(),
   totalTokenCount: import_v47.z.number().nullish(),
   // https://cloud.google.com/vertex-ai/generative-ai/docs/reference/rest/v1/GenerateContentResponse#TrafficType
   trafficType: import_v47.z.string().nullish(),
   serviceTier: import_v47.z.string().nullish(),
   // https://ai.google.dev/api/generate-content#Modality
   promptTokensDetails: tokenDetailsSchema,
-  candidatesTokensDetails: tokenDetailsSchema
-});
+  cacheTokensDetails: tokenDetailsSchema,
+  candidatesTokensDetails: tokenDetailsSchema,
+  toolUsePromptTokensDetails: tokenDetailsSchema
+}).catchall(import_v47.z.json());
 var getUrlContextMetadataSchema = () => import_v47.z.object({
   urlMetadata: import_v47.z.array(
     import_v47.z.object({
@@ -5238,8 +5241,8 @@ var googleInteractionsLanguageModelOptions = (0, import_provider_utils19.lazySch
       thinkingSummaries: import_v418.z.enum(["auto", "none"]).nullish(),
       /**
        * Output-format entries that map directly to the API's `response_format`
-       * array. Use this to request image, audio, or non-JSON text outputs
-       * with full control over `mime_type`, `aspect_ratio`, and `image_size`.
+       * array. Use this to request image, audio, video, or non-JSON text
+       * outputs with modality-specific controls.
        *
        * Entries are sent in order. The AI SDK call-level `responseFormat: {
        * type: 'json', schema }` still drives JSON-mode and adds a matching
@@ -5276,6 +5279,14 @@ var googleInteractionsLanguageModelOptions = (0, import_provider_utils19.lazySch
           import_v418.z.object({
             type: import_v418.z.literal("audio"),
             mimeType: import_v418.z.string().nullish()
+          }).loose(),
+          import_v418.z.object({
+            type: import_v418.z.literal("video"),
+            aspectRatio: import_v418.z.enum(["16:9", "9:16"]).nullish(),
+            resolution: import_v418.z.enum(["360p", "720p", "1080p", "4k"]).nullish(),
+            duration: import_v418.z.string().nullish(),
+            delivery: import_v418.z.enum(["inline", "uri"]).nullish(),
+            gcsUri: import_v418.z.string().nullish()
           }).loose()
         ])
       ).nullish(),
@@ -6162,7 +6173,7 @@ var GoogleInteractionsLanguageModel = class {
     };
   }
   async getArgs(options) {
-    var _a, _b, _c, _d, _e, _f, _g, _h, _i, _j, _k, _l, _m, _n, _o, _p, _q, _r, _s, _t, _u, _v, _w, _x, _y, _z, _A;
+    var _a, _b, _c, _d, _e, _f, _g, _h, _i, _j, _k, _l, _m, _n, _o, _p, _q, _r, _s, _t, _u, _v, _w, _x, _y, _z, _A, _B, _C, _D, _E, _F;
     const warnings = [];
     const opts = await (0, import_provider_utils23.parseProviderOptions)({
       provider: "google",
@@ -6238,6 +6249,17 @@ var GoogleInteractionsLanguageModel = class {
               mime_type: (_g = entry.mimeType) != null ? _g : void 0
             })
           );
+        } else if (entry.type === "video") {
+          responseFormatEntries.push(
+            pruneUndefined({
+              type: "video",
+              aspect_ratio: (_h = entry.aspectRatio) != null ? _h : void 0,
+              resolution: (_i = entry.resolution) != null ? _i : void 0,
+              duration: (_j = entry.duration) != null ? _j : void 0,
+              delivery: (_k = entry.delivery) != null ? _k : void 0,
+              gcs_uri: (_l = entry.gcsUri) != null ? _l : void 0
+            })
+          );
         }
       }
     }
@@ -6247,13 +6269,13 @@ var GoogleInteractionsLanguageModel = class {
       warnings: convWarnings
     } = convertToGoogleInteractionsInput({
       prompt: options.prompt,
-      previousInteractionId: (_h = opts == null ? void 0 : opts.previousInteractionId) != null ? _h : void 0,
-      store: (_i = opts == null ? void 0 : opts.store) != null ? _i : void 0,
-      mediaResolution: (_j = opts == null ? void 0 : opts.mediaResolution) != null ? _j : void 0
+      previousInteractionId: (_m = opts == null ? void 0 : opts.previousInteractionId) != null ? _m : void 0,
+      store: (_n = opts == null ? void 0 : opts.store) != null ? _n : void 0,
+      mediaResolution: (_o = opts == null ? void 0 : opts.mediaResolution) != null ? _o : void 0
     });
     warnings.push(...convWarnings);
     let systemInstruction = convertedSystemInstruction;
-    const optionSystemInstruction = (_k = opts == null ? void 0 : opts.systemInstruction) != null ? _k : void 0;
+    const optionSystemInstruction = (_p = opts == null ? void 0 : opts.systemInstruction) != null ? _p : void 0;
     if (systemInstruction != null && optionSystemInstruction != null) {
       warnings.push({
         type: "other",
@@ -6292,14 +6314,14 @@ var GoogleInteractionsLanguageModel = class {
       generationConfig = void 0;
     } else {
       generationConfig = pruneUndefined({
-        temperature: (_l = options.temperature) != null ? _l : void 0,
-        top_p: (_m = options.topP) != null ? _m : void 0,
-        top_k: (_n = options.topK) != null ? _n : void 0,
-        seed: (_o = options.seed) != null ? _o : void 0,
+        temperature: (_q = options.temperature) != null ? _q : void 0,
+        top_p: (_r = options.topP) != null ? _r : void 0,
+        top_k: (_s = options.topK) != null ? _s : void 0,
+        seed: (_t = options.seed) != null ? _t : void 0,
         stop_sequences: options.stopSequences != null && options.stopSequences.length > 0 ? options.stopSequences : void 0,
-        max_output_tokens: (_p = options.maxOutputTokens) != null ? _p : void 0,
-        thinking_level: (_q = opts == null ? void 0 : opts.thinkingLevel) != null ? _q : void 0,
-        thinking_summaries: (_r = opts == null ? void 0 : opts.thinkingSummaries) != null ? _r : void 0,
+        max_output_tokens: (_u = options.maxOutputTokens) != null ? _u : void 0,
+        thinking_level: (_v = opts == null ? void 0 : opts.thinkingLevel) != null ? _v : void 0,
+        thinking_summaries: (_w = opts == null ? void 0 : opts.thinkingSummaries) != null ? _w : void 0,
         tool_choice: toolChoiceForBody
       });
       if ((opts == null ? void 0 : opts.imageConfig) != null) {
@@ -6326,9 +6348,9 @@ var GoogleInteractionsLanguageModel = class {
       if (ac.type === "deep-research") {
         agentConfig = pruneUndefined({
           type: "deep-research",
-          thinking_summaries: (_s = ac.thinkingSummaries) != null ? _s : void 0,
-          visualization: (_t = ac.visualization) != null ? _t : void 0,
-          collaborative_planning: (_u = ac.collaborativePlanning) != null ? _u : void 0
+          thinking_summaries: (_x = ac.thinkingSummaries) != null ? _x : void 0,
+          visualization: (_y = ac.visualization) != null ? _y : void 0,
+          collaborative_planning: (_z = ac.collaborativePlanning) != null ? _z : void 0
         });
       } else if (ac.type === "dynamic") {
         agentConfig = { type: "dynamic" };
@@ -6345,7 +6367,7 @@ var GoogleInteractionsLanguageModel = class {
         environment = opts.environment;
       } else {
         const env = opts.environment;
-        const sources = (_v = env.sources) == null ? void 0 : _v.map((s) => {
+        const sources = (_A = env.sources) == null ? void 0 : _A.map((s) => {
           var _a2;
           if (s.type === "inline") {
             return {
@@ -6390,20 +6412,20 @@ var GoogleInteractionsLanguageModel = class {
       tools: toolsForBody,
       response_format: responseFormatEntries.length > 0 ? responseFormatEntries : void 0,
       response_modalities: (opts == null ? void 0 : opts.responseModalities) != null ? opts.responseModalities : void 0,
-      previous_interaction_id: (_w = opts == null ? void 0 : opts.previousInteractionId) != null ? _w : void 0,
-      service_tier: (_x = opts == null ? void 0 : opts.serviceTier) != null ? _x : void 0,
-      store: (_y = opts == null ? void 0 : opts.store) != null ? _y : void 0,
+      previous_interaction_id: (_B = opts == null ? void 0 : opts.previousInteractionId) != null ? _B : void 0,
+      service_tier: (_C = opts == null ? void 0 : opts.serviceTier) != null ? _C : void 0,
+      store: (_D = opts == null ? void 0 : opts.store) != null ? _D : void 0,
       generation_config: generationConfig != null && Object.keys(generationConfig).length > 0 ? generationConfig : void 0,
       agent_config: agentConfig,
       environment,
-      background: (_z = opts == null ? void 0 : opts.background) != null ? _z : void 0
+      background: (_E = opts == null ? void 0 : opts.background) != null ? _E : void 0
     });
     return {
       args,
       warnings,
       isAgent,
       isBackground: (opts == null ? void 0 : opts.background) === true,
-      pollingTimeoutMs: (_A = opts == null ? void 0 : opts.pollingTimeoutMs) != null ? _A : void 0
+      pollingTimeoutMs: (_F = opts == null ? void 0 : opts.pollingTimeoutMs) != null ? _F : void 0
     };
   }
   async doGenerate(options) {