@ai-sdk/xai 3.0.120 → 3.0.122

package/dist/index.mjs

@@ -2988,7 +2988,7 @@ var xaiTools = {
 };
 
 // src/version.ts
-var VERSION = true ? "3.0.120" : "0.0.0-test";
+var VERSION = true ? "3.0.122" : "0.0.0-test";
 
 // src/xai-video-model.ts
 import {
@@ -3012,7 +3012,7 @@ import { z as z17 } from "zod/v4";
 import { lazySchema as lazySchema5, zodSchema as zodSchema5 } from "@ai-sdk/provider-utils";
 import { z as z16 } from "zod/v4";
 var nonEmptyStringSchema = z16.string().min(1);
-var resolutionSchema = z16.enum(["480p", "720p"]);
+var resolutionSchema = z16.enum(["480p", "720p", "1080p"]);
 var modeSchema = z16.enum(["edit-video", "extend-video", "reference-to-video"]);
 var baseFields = {
   pollIntervalMs: z16.number().positive().nullish(),
@@ -3022,22 +3022,28 @@ var baseFields = {
 var userField = {
   user: z16.string().optional()
 };
+var referenceVoiceIdsField = {
+  referenceVoiceIds: z16.array(nonEmptyStringSchema).max(3).optional()
+};
 var editVideoSchema = z16.object({
   ...baseFields,
   ...userField,
   mode: z16.literal("edit-video"),
   videoUrl: nonEmptyStringSchema,
-  referenceImageUrls: z16.undefined().optional()
+  referenceImageUrls: z16.undefined().optional(),
+  referenceVoiceIds: z16.undefined().optional()
 });
 var extendVideoSchema = z16.object({
   ...baseFields,
   mode: z16.literal("extend-video"),
   videoUrl: nonEmptyStringSchema,
-  referenceImageUrls: z16.undefined().optional()
+  referenceImageUrls: z16.undefined().optional(),
+  referenceVoiceIds: z16.undefined().optional()
 });
 var referenceToVideoSchema = z16.object({
   ...baseFields,
   ...userField,
+  ...referenceVoiceIdsField,
   mode: z16.literal("reference-to-video"),
   referenceImageUrls: z16.array(nonEmptyStringSchema).min(1).max(7),
   videoUrl: z16.undefined().optional()
@@ -3045,6 +3051,7 @@ var referenceToVideoSchema = z16.object({
 var autoDetectSchema = z16.object({
   ...baseFields,
   ...userField,
+  ...referenceVoiceIdsField,
   mode: z16.undefined().optional(),
   videoUrl: nonEmptyStringSchema.optional(),
   referenceImageUrls: z16.array(nonEmptyStringSchema).min(1).max(7).optional()
@@ -3059,6 +3066,7 @@ var runtimeSchema = z16.object({
   mode: modeSchema.optional(),
   videoUrl: nonEmptyStringSchema.optional(),
   referenceImageUrls: z16.array(nonEmptyStringSchema).min(1).max(7).optional(),
+  referenceVoiceIds: z16.array(nonEmptyStringSchema).max(3).optional(),
   user: z16.string().optional(),
   ...baseFields
 }).passthrough();
@@ -3068,6 +3076,7 @@ var xaiVideoModelOptionsSchema = lazySchema5(
 
 // src/xai-video-model.ts
 var RESOLUTION_MAP = {
+  "1920x1080": "1080p",
   "1280x720": "720p",
   "854x480": "480p",
   "640x480": "480p"
@@ -3089,6 +3098,7 @@ 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;
@@ -3098,26 +3108,29 @@ 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;
@@ -3126,9 +3139,8 @@ 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;
@@ -3235,7 +3247,7 @@ 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.`
         });
       }
     }
@@ -3274,13 +3286,47 @@ 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) {
@@ -3295,6 +3341,7 @@ var XaiVideoModel = class {
           "resolution",
           "videoUrl",
           "referenceImageUrls",
+          "referenceVoiceIds",
           "user"
         ].includes(key)) {
           body[key] = value;