@picsart/ai-sdk 6.45.0 → 6.47.0

package/index.d.ts

@@ -141,6 +141,12 @@ type ModelInputById = {
         turboMode?: boolean;
         seed?: number;
     };
+    "bytedance-omnihuman-v2.0": {
+        prompt?: string;
+        imageUrls: [string, ...string[]];
+        audioUrl: string;
+        enableWatermark?: boolean;
+    };
     "bytedance-video-enhance": {
         videoUrl: string;
         quality?: "standard" | "professional";
@@ -3558,6 +3564,7 @@ declare function createClient(config: ClientConfig): AiClient;
 declare const Models: {
     readonly AsyncFlashV1: "async-flash-v1";
     readonly BytedanceOmnihumanV15: "bytedance-omnihuman-v1.5";
+    readonly BytedanceOmnihumanV20: "bytedance-omnihuman-v2.0";
     readonly BytedanceVideoEnhance: "bytedance-video-enhance";
     readonly BytedanceVideoUpscaler: "bytedance-video-upscaler";
     readonly CaptionsaiVideoCaptions: "captionsai-video-captions";

package/index.js

@@ -3009,6 +3009,36 @@ var { MODELS: MODELS6 } = defineModels("bytedance", [
       ...p.range("seed", -1, 2147483647, -1)
     }
   },
+  {
+    id: "bytedance-omnihuman-v2.0",
+    name: "ByteDance OmniHuman 2.0",
+    addedAt: "2026-10-16",
+    workflow: "bytedance/omnihuman/v2.0",
+    // Pricing key of the direct BytePlus Vision AI integration. The catalog `id`
+    // keeps its `v2.0` spelling for stability, so the backend key is spelled out.
+    // Priced per second of output video (equal to the driving audio length).
+    modelId: "bytedance-omnihuman-2.0",
+    // Output video length equals the driving audio (4..300s), so wall time scales
+    // with the clip. Representative mid-length figure, mirroring v1.5 — not a ceiling.
+    estimatedTime: 250,
+    mode: "video",
+    inputType: "i2v",
+    description: "Animate one or more subjects with realistic body movement driven by audio.",
+    features: [
+      feat("Image Input", "input"),
+      feat("Audio Input", "audio"),
+      feat("Multi-Subject", "quality")
+    ],
+    paramConfig: {
+      // Optional prompt steers scene/movement/camera, NOT lip-sync (the audio does).
+      ...params.prompt({ required: false }),
+      // v2.0 takes an array of source images (person, pet or character).
+      ...params.imageInput(4, "Source Image(s)", true),
+      ...params.audioInput("Audio Track", true),
+      // Vendor default is watermark-on; pass false for an un-watermarked result.
+      ...p.boolean("enableWatermark", true, "Watermark")
+    }
+  },
   {
     id: "bytedance-video-enhance",
     name: "ByteDance Video Enhance",
@@ -3070,9 +3100,16 @@ var buildBytedanceOmniHumanPayload = (input) => ({
   ...input.turboMode ? { turbo_mode: true } : {},
   ...input.seed != null && input.seed !== -1 ? { seed: input.seed } : {}
 });
+var buildBytedanceOmniHumanV2Payload = (input) => ({
+  image_urls: input.imageUrls,
+  audio_url: input.audioUrl,
+  enable_watermark: input.enableWatermark,
+  ...input.prompt ? { prompt: input.prompt } : {}
+});
 registerPayloads(MODELS6, {
   "bytedance-video-enhance": buildBytedanceVideoEnhancePayload,
-  "bytedance-omnihuman-v1.5": buildBytedanceOmniHumanPayload
+  "bytedance-omnihuman-v1.5": buildBytedanceOmniHumanPayload,
+  "bytedance-omnihuman-v2.0": buildBytedanceOmniHumanV2Payload
 });
 
 // src/vendors/catalog/videography.ts
@@ -15222,6 +15259,7 @@ function createClient(config) {
 // src/generated/model-constants.ts
 var AsyncFlashV1 = "async-flash-v1";
 var BytedanceOmnihumanV15 = "bytedance-omnihuman-v1.5";
+var BytedanceOmnihumanV20 = "bytedance-omnihuman-v2.0";
 var BytedanceVideoEnhance = "bytedance-video-enhance";
 var BytedanceVideoUpscaler = "bytedance-video-upscaler";
 var CaptionsaiVideoCaptions = "captionsai-video-captions";
@@ -15495,6 +15533,7 @@ var Wan30VideoPrime = "wan-3.0-video-prime";
 var Models = {
   AsyncFlashV1,
   BytedanceOmnihumanV15,
+  BytedanceOmnihumanV20,
   BytedanceVideoEnhance,
   BytedanceVideoUpscaler,
   CaptionsaiVideoCaptions,

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@picsart/ai-sdk",
-  "version": "6.45.0",
+  "version": "6.47.0",
   "type": "module",
   "description": "Type-safe SDK for 100+ AI models — image, video, audio, and text generation with Picsart",
   "license": "MIT",