@ai-sdk/google 4.0.16 → 4.0.18

package/docs/15-google.mdx

@@ -498,7 +498,7 @@ The response will contain the tool calls and results from the code execution.
 With [Google Search grounding](https://ai.google.dev/gemini-api/docs/google-search),
 the model has access to the latest information using Google Search.
 
-```ts highlight="8,17-20"
+```ts highlight="8,17-19"
 import { google } from '@ai-sdk/google';
 import { GoogleProviderMetadata } from '@ai-sdk/google';
 import { generateText } from 'ai';
@@ -657,7 +657,7 @@ Google provides a provider-defined URL context tool.
 
 The URL context tool allows you to provide specific URLs that you want the model to analyze directly in from the prompt.
 
-```ts highlight="9,13-17"
+```ts highlight="9,13-15"
 import { google } from '@ai-sdk/google';
 import { generateText } from 'ai';
 
@@ -818,7 +818,7 @@ This enables the model to provide answers based on your specific data sources an
   Google provider (`@ai-sdk/google`) to use this feature.
 </Note>
 
-```ts highlight="8,17-20"
+```ts highlight="5-8,13-17,25-27"
 import { createGoogleVertex } from '@ai-sdk/google-vertex';
 import { GoogleProviderMetadata } from '@ai-sdk/google';
 import { generateText } from 'ai';
@@ -2024,7 +2024,7 @@ spoken text, so `instructions: 'Say cheerfully'` with `text: 'Hello'` speaks `Sa
 For multi-speaker dialogue, pass a `multiSpeakerVoiceConfig` through `providerOptions`. Each speaker
 name must match a name used in the input text. When set, it overrides the top-level `voice`.
 
-```ts highlight="8-23"
+```ts highlight="7-22"
 import { generateSpeech } from 'ai';
 import { google, type GoogleSpeechModelOptions } from '@ai-sdk/google';
 

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@ai-sdk/google",
-  "version": "4.0.16",
+  "version": "4.0.18",
   "type": "module",
   "license": "Apache-2.0",
   "sideEffects": false,
@@ -36,7 +36,7 @@
   },
   "dependencies": {
     "@ai-sdk/provider": "4.0.3",
-    "@ai-sdk/provider-utils": "5.0.10"
+    "@ai-sdk/provider-utils": "5.0.11"
   },
   "devDependencies": {
     "@types/node": "22.19.19",

package/src/google-language-model-options.ts

@@ -176,6 +176,46 @@ export const googleLanguageModelOptions = lazySchema(() =>
             ])
             .optional(),
           imageSize: z.enum(['1K', '2K', '4K', '512']).optional(),
+
+          /**
+           * Optional. Controls the generation of people in images.
+           * Vertex AI only.
+           */
+          personGeneration: z
+            .enum([
+              'PERSON_GENERATION_UNSPECIFIED',
+              'ALLOW_ALL',
+              'ALLOW_ADULT',
+              'ALLOW_NONE',
+            ])
+            .optional(),
+
+          /**
+           * Optional. Controls whether generation of prominent people
+           * (celebrities) is allowed. When set together with
+           * `personGeneration`, `personGeneration` takes precedence.
+           * Vertex AI only.
+           *
+           * https://docs.cloud.google.com/vertex-ai/generative-ai/docs/reference/rest/v1/GenerationConfig
+           */
+          prominentPeople: z
+            .enum([
+              'PROMINENT_PEOPLE_UNSPECIFIED',
+              'ALLOW_PROMINENT_PEOPLE',
+              'BLOCK_PROMINENT_PEOPLE',
+            ])
+            .optional(),
+
+          /**
+           * Optional. The image output format for generated images.
+           * Vertex AI only.
+           */
+          imageOutputOptions: z
+            .object({
+              mimeType: z.enum(['image/jpeg', 'image/png']).optional(),
+              compressionQuality: z.number().optional(),
+            })
+            .optional(),
         })
         .optional(),
 

package/src/google-language-model.ts

@@ -227,6 +227,35 @@ export class GoogleLanguageModel implements LanguageModelV4 {
       ? undefined
       : googleOptions?.serviceTier;
 
+    // personGeneration, prominentPeople and imageOutputOptions are only
+    // supported by the Vertex AI API, the Gemini API rejects them.
+    let imageConfig = googleOptions?.imageConfig;
+    if (imageConfig != null && !isVertexProvider) {
+      const {
+        personGeneration,
+        prominentPeople,
+        imageOutputOptions,
+        ...geminiApiImageConfig
+      } = imageConfig;
+      const droppedImageConfigFields = Object.entries({
+        personGeneration,
+        prominentPeople,
+        imageOutputOptions,
+      })
+        .filter(([, value]) => value != null)
+        .map(([key]) => `'imageConfig.${key}'`);
+      if (droppedImageConfigFields.length > 0) {
+        warnings.push({
+          type: 'other',
+          message:
+            `${droppedImageConfigFields.join(', ')} ` +
+            `${droppedImageConfigFields.length === 1 ? 'is a Vertex AI option and is' : 'are Vertex AI options and are'} ` +
+            `ignored with the current Google provider (${this.config.provider}).`,
+        });
+        imageConfig = geminiApiImageConfig;
+      }
+    }
+
     const isGemmaModel = this.modelId.toLowerCase().startsWith('gemma-');
     const isGemini3Model = /^gemini-3[.-]/.test(this.modelId);
     const supportsFunctionResponseParts = isGemini3Model;
@@ -328,9 +357,7 @@ export class GoogleLanguageModel implements LanguageModelV4 {
           ...(googleOptions?.mediaResolution && {
             mediaResolution: googleOptions.mediaResolution,
           }),
-          ...(googleOptions?.imageConfig && {
-            imageConfig: googleOptions.imageConfig,
-          }),
+          ...(imageConfig && { imageConfig }),
         },
         contents,
         systemInstruction: isGemmaModel ? undefined : systemInstruction,