@ai-sdk/xai 3.0.120 → 3.0.121

package/CHANGELOG.md

@@ -1,5 +1,12 @@
 # @ai-sdk/xai
 
+## 3.0.121
+
+### Patch Changes
+
+- 70a6c3b: feat (provider/xai): support Grok Imagine Video 1.5. Adds the `grok-imagine-video-1.5` model id and native `1080p` for text-to-video and image-to-video (the standard `resolution: '1920x1080'` now maps to `1080p`). Reference-to-video remains capped at `720p`, so a `1080p` request in that mode is downgraded with a warning. Also fixes reference routing: previously any non-empty `inputReferences` array selected reference-to-video, so an array holding only a non-image reference sent `reference_images: []` with no usable references.
+- 70a6c3b: feat (provider/xai): add `referenceVoiceIds` for reference-to-video reference audio. Pass up to 3 xAI preset voice ids (e.g. `['eve']`) and reference them from the prompt with `<AUDIO_0>`–`<AUDIO_2>`; they are sent as `reference_audios: [{ voice_id }]` on `POST /v1/videos/generations`.
+
 ## 3.0.120
 
 ### Patch Changes

package/dist/index.d.mts

@@ -120,11 +120,12 @@ declare const xaiImageModelOptions: z.ZodObject<{
 }, z.core.$strip>;
 type XaiImageModelOptions = z.infer<typeof xaiImageModelOptions>;
 
-type XaiVideoModelId = 'grok-imagine-video' | (string & {});
+type XaiVideoModelId = 'grok-imagine-video' | 'grok-imagine-video-1.5' | (string & {});
 
 declare const resolutionSchema: z.ZodEnum<{
     "480p": "480p";
     "720p": "720p";
+    "1080p": "1080p";
 }>;
 type XaiVideoResolution = z.infer<typeof resolutionSchema>;
 interface XaiVideoSharedOptions {
@@ -161,6 +162,10 @@ interface XaiVideoReferenceToVideoOptions extends XaiVideoSharedOptions, XaiVide
     mode: 'reference-to-video';
     /** Reference image URLs (1-7) for R2V generation. */
     referenceImageUrls: string[];
+    /**
+     * Preset voice ids (up to 3) that give the subject a voice.
+     */
+    referenceVoiceIds?: string[];
 }
 interface XaiVideoGenerationOptions extends XaiVideoSharedOptions, XaiVideoUserOptions {
     mode?: undefined;
@@ -182,6 +187,10 @@ interface XaiLegacyReferenceToVideoOptions extends XaiVideoSharedOptions, XaiVid
      */
     mode?: undefined;
     referenceImageUrls: string[];
+    /**
+     * Preset voice ids (up to 3) that give the subject a voice.
+     */
+    referenceVoiceIds?: string[];
 }
 /**
  * Provider options for xAI video generation.
@@ -193,6 +202,9 @@ interface XaiLegacyReferenceToVideoOptions extends XaiVideoSharedOptions, XaiVid
  * - `'reference-to-video'` + `referenceImageUrls` -- R2V generation  (`POST /v1/videos/generations`)
  * - no `mode`                                     -- standard generation from text prompts or image input
  *
+ * Reference images may also come from the top-level `inputReferences` option
+ * instead of `referenceImageUrls`.
+ *
  * Runtime remains backward compatible with legacy auto-detected provider
  * options, but the public TypeScript type is intentionally explicit so editors
  * can suggest valid modes and flag invalid field combinations.

package/dist/index.d.ts

@@ -120,11 +120,12 @@ declare const xaiImageModelOptions: z.ZodObject<{
 }, z.core.$strip>;
 type XaiImageModelOptions = z.infer<typeof xaiImageModelOptions>;
 
-type XaiVideoModelId = 'grok-imagine-video' | (string & {});
+type XaiVideoModelId = 'grok-imagine-video' | 'grok-imagine-video-1.5' | (string & {});
 
 declare const resolutionSchema: z.ZodEnum<{
     "480p": "480p";
     "720p": "720p";
+    "1080p": "1080p";
 }>;
 type XaiVideoResolution = z.infer<typeof resolutionSchema>;
 interface XaiVideoSharedOptions {
@@ -161,6 +162,10 @@ interface XaiVideoReferenceToVideoOptions extends XaiVideoSharedOptions, XaiVide
     mode: 'reference-to-video';
     /** Reference image URLs (1-7) for R2V generation. */
     referenceImageUrls: string[];
+    /**
+     * Preset voice ids (up to 3) that give the subject a voice.
+     */
+    referenceVoiceIds?: string[];
 }
 interface XaiVideoGenerationOptions extends XaiVideoSharedOptions, XaiVideoUserOptions {
     mode?: undefined;
@@ -182,6 +187,10 @@ interface XaiLegacyReferenceToVideoOptions extends XaiVideoSharedOptions, XaiVid
      */
     mode?: undefined;
     referenceImageUrls: string[];
+    /**
+     * Preset voice ids (up to 3) that give the subject a voice.
+     */
+    referenceVoiceIds?: string[];
 }
 /**
  * Provider options for xAI video generation.
@@ -193,6 +202,9 @@ interface XaiLegacyReferenceToVideoOptions extends XaiVideoSharedOptions, XaiVid
  * - `'reference-to-video'` + `referenceImageUrls` -- R2V generation  (`POST /v1/videos/generations`)
  * - no `mode`                                     -- standard generation from text prompts or image input
  *
+ * Reference images may also come from the top-level `inputReferences` option
+ * instead of `referenceImageUrls`.
+ *
  * Runtime remains backward compatible with legacy auto-detected provider
  * options, but the public TypeScript type is intentionally explicit so editors
  * can suggest valid modes and flag invalid field combinations.

package/dist/index.js

@@ -2967,7 +2967,7 @@ var xaiTools = {
 };
 
 // src/version.ts
-var VERSION = true ? "3.0.120" : "0.0.0-test";
+var VERSION = true ? "3.0.121" : "0.0.0-test";
 
 // src/xai-video-model.ts
 var import_provider6 = require("@ai-sdk/provider");
@@ -2978,7 +2978,7 @@ var import_v417 = require("zod/v4");
 var import_provider_utils15 = require("@ai-sdk/provider-utils");
 var import_v416 = require("zod/v4");
 var nonEmptyStringSchema = import_v416.z.string().min(1);
-var resolutionSchema = import_v416.z.enum(["480p", "720p"]);
+var resolutionSchema = import_v416.z.enum(["480p", "720p", "1080p"]);
 var modeSchema = import_v416.z.enum(["edit-video", "extend-video", "reference-to-video"]);
 var baseFields = {
   pollIntervalMs: import_v416.z.number().positive().nullish(),
@@ -2988,22 +2988,28 @@ var baseFields = {
 var userField = {
   user: import_v416.z.string().optional()
 };
+var referenceVoiceIdsField = {
+  referenceVoiceIds: import_v416.z.array(nonEmptyStringSchema).max(3).optional()
+};
 var editVideoSchema = import_v416.z.object({
   ...baseFields,
   ...userField,
   mode: import_v416.z.literal("edit-video"),
   videoUrl: nonEmptyStringSchema,
-  referenceImageUrls: import_v416.z.undefined().optional()
+  referenceImageUrls: import_v416.z.undefined().optional(),
+  referenceVoiceIds: import_v416.z.undefined().optional()
 });
 var extendVideoSchema = import_v416.z.object({
   ...baseFields,
   mode: import_v416.z.literal("extend-video"),
   videoUrl: nonEmptyStringSchema,
-  referenceImageUrls: import_v416.z.undefined().optional()
+  referenceImageUrls: import_v416.z.undefined().optional(),
+  referenceVoiceIds: import_v416.z.undefined().optional()
 });
 var referenceToVideoSchema = import_v416.z.object({
   ...baseFields,
   ...userField,
+  ...referenceVoiceIdsField,
   mode: import_v416.z.literal("reference-to-video"),
   referenceImageUrls: import_v416.z.array(nonEmptyStringSchema).min(1).max(7),
   videoUrl: import_v416.z.undefined().optional()
@@ -3011,6 +3017,7 @@ var referenceToVideoSchema = import_v416.z.object({
 var autoDetectSchema = import_v416.z.object({
   ...baseFields,
   ...userField,
+  ...referenceVoiceIdsField,
   mode: import_v416.z.undefined().optional(),
   videoUrl: nonEmptyStringSchema.optional(),
   referenceImageUrls: import_v416.z.array(nonEmptyStringSchema).min(1).max(7).optional()
@@ -3025,6 +3032,7 @@ var runtimeSchema = import_v416.z.object({
   mode: modeSchema.optional(),
   videoUrl: nonEmptyStringSchema.optional(),
   referenceImageUrls: import_v416.z.array(nonEmptyStringSchema).min(1).max(7).optional(),
+  referenceVoiceIds: import_v416.z.array(nonEmptyStringSchema).max(3).optional(),
   user: import_v416.z.string().optional(),
   ...baseFields
 }).passthrough();
@@ -3034,6 +3042,7 @@ var xaiVideoModelOptionsSchema = (0, import_provider_utils15.lazySchema)(
 
 // src/xai-video-model.ts
 var RESOLUTION_MAP = {
+  "1920x1080": "1080p",
   "1280x720": "720p",
   "854x480": "480p",
   "640x480": "480p"
@@ -3055,6 +3064,7 @@ function getTopLevelMediaType(mediaType) {
   return slashIndex === -1 ? mediaType : mediaType.substring(0, slashIndex);
 }
 var isVideoFile = (file) => file.mediaType != null && getTopLevelMediaType(file.mediaType) === "video";
+var isImageReference = (file) => file.mediaType == null || getTopLevelMediaType(file.mediaType) === "image";
 function fileToXaiImageUrl(file) {
   if (file.type === "url") {
     return file.url;
@@ -3064,26 +3074,29 @@ function fileToXaiImageUrl(file) {
 }
 function resolveReferenceImages(options, xaiOptions, warnings) {
   if (options.inputReferences != null && options.inputReferences.length > 0) {
-    const imageReferences = options.inputReferences.filter((reference) => {
-      if (isVideoFile(reference)) {
+    const imageFiles = [];
+    for (const reference of options.inputReferences) {
+      if (!isImageReference(reference)) {
         warnings.push({
           type: "unsupported",
           feature: "inputReferences",
-          details: 'xAI reference-to-video accepts image references only. The video reference was ignored. Use providerOptions.xai.mode "extend-video" to continue from a video.'
+          details: isVideoFile(reference) ? 'xAI reference-to-video accepts image references only. The video reference was ignored. Use providerOptions.xai.mode "extend-video" to continue from a video.' : "xAI reference-to-video accepts image references only. The non-image reference was ignored."
         });
-        return false;
+        continue;
       }
-      return true;
-    });
-    return imageReferences.map((reference) => ({
-      url: fileToXaiImageUrl(reference)
-    }));
+      imageFiles.push(reference);
+    }
+    return imageFiles.length > 0 ? imageFiles.map((reference) => ({ url: fileToXaiImageUrl(reference) })) : void 0;
   }
   if ((xaiOptions == null ? void 0 : xaiOptions.referenceImageUrls) != null && xaiOptions.referenceImageUrls.length > 0) {
     return xaiOptions.referenceImageUrls.map((url) => ({ url }));
   }
   return void 0;
 }
+function hasImageInputReference(options) {
+  var _a, _b;
+  return (_b = (_a = options.inputReferences) == null ? void 0 : _a.some(isImageReference)) != null ? _b : false;
+}
 function resolveVideoMode(options, xaiOptions) {
   if ((xaiOptions == null ? void 0 : xaiOptions.mode) != null) {
     return xaiOptions.mode;
@@ -3092,9 +3105,8 @@ function resolveVideoMode(options, xaiOptions) {
     return "edit-video";
   }
   const hasFrameImages = options.frameImages != null && options.frameImages.length > 0;
-  const hasInputReferences = options.inputReferences != null && options.inputReferences.length > 0;
   const hasLegacyReferenceUrls = (xaiOptions == null ? void 0 : xaiOptions.referenceImageUrls) != null && xaiOptions.referenceImageUrls.length > 0;
-  if (!hasFrameImages && (hasInputReferences || hasLegacyReferenceUrls)) {
+  if (!hasFrameImages && (hasImageInputReference(options) || hasLegacyReferenceUrls)) {
     return "reference-to-video";
   }
   return void 0;
@@ -3201,7 +3213,7 @@ var XaiVideoModel = class {
         warnings.push({
           type: "unsupported",
           feature: "resolution",
-          details: `Unrecognized resolution "${options.resolution}". Use providerOptions.xai.resolution with "480p" or "720p" instead.`
+          details: `Unrecognized resolution "${options.resolution}". Use providerOptions.xai.resolution with "480p", "720p", or "1080p" instead.`
         });
       }
     }
@@ -3240,13 +3252,47 @@ var XaiVideoModel = class {
       );
       if (referenceImages != null) {
         body.reference_images = referenceImages;
+      } else {
+        warnings.push({
+          type: "unsupported",
+          feature: "referenceImages",
+          details: "xAI reference-to-video requires at least one image reference. The video will be generated without reference images."
+        });
       }
+      const referenceVoiceIds = xaiOptions == null ? void 0 : xaiOptions.referenceVoiceIds;
+      if (referenceVoiceIds != null && referenceVoiceIds.length > 0) {
+        body.reference_audios = referenceVoiceIds.map((voiceId) => ({
+          voice_id: voiceId
+        }));
+      }
+      if (body.resolution === "1080p") {
+        warnings.push({
+          type: "unsupported",
+          feature: "resolution",
+          details: "xAI reference-to-video is limited to 720p. The request was downgraded from 1080p to 720p."
+        });
+        body.resolution = "720p";
+      }
+    }
+    if (body.resolution === "1080p" && this.modelId === "grok-imagine-video") {
+      warnings.push({
+        type: "unsupported",
+        feature: "resolution",
+        details: 'xAI model "grok-imagine-video" does not support 1080p. Use "grok-imagine-video-1.5" for 1080p, or a lower resolution. The request was sent with 1080p.'
+      });
     }
     if (options.inputReferences != null && options.inputReferences.length > 0 && !hasReferenceImages) {
       warnings.push({
         type: "unsupported",
         feature: "inputReferences",
-        details: "xAI only supports inputReferences for reference-to-video generation. The reference images were ignored."
+        details: hasImageInputReference(options) ? "xAI only supports inputReferences for reference-to-video generation. The reference images were ignored." : "xAI reference-to-video requires at least one image reference. The references were ignored."
+      });
+    }
+    if ((xaiOptions == null ? void 0 : xaiOptions.referenceVoiceIds) != null && xaiOptions.referenceVoiceIds.length > 0 && !hasReferenceImages) {
+      warnings.push({
+        type: "unsupported",
+        feature: "referenceVoiceIds",
+        details: "xAI only supports reference voices for reference-to-video generation. The reference voices were ignored."
       });
     }
     if (!isExtension && (xaiOptions == null ? void 0 : xaiOptions.user) !== void 0) {
@@ -3261,6 +3307,7 @@ var XaiVideoModel = class {
           "resolution",
           "videoUrl",
           "referenceImageUrls",
+          "referenceVoiceIds",
           "user"
         ].includes(key)) {
           body[key] = value;