vidspotai-shared 1.0.124 → 1.0.126

package/lib/globals/aiModels/providers/alibaba.js

@@ -1,6 +1,8 @@
 "use strict";
 Object.defineProperty(exports, "__esModule", { value: true });
 exports.alibabaConfigs = void 0;
+/** Wan 2.6 / 2.7: any integer 2–15s (re-read 2026-09-29). */
+const WAN_2_15_DURATIONS = Array.from({ length: 14 }, (_, i) => i + 2);
 const wanCommonFields = {
     prompt: { required: true },
     aspectRatio: { allowedValues: ["16:9", "9:16", "1:1", "4:3", "3:4"] },
@@ -11,7 +13,233 @@ const wanCommonFields = {
     watermark: {},
     promptOptimizer: {},
 };
+// ───────────────────────────────────────────────────────────────────────
+// Wan 3.0 + HappyHorse — the input.media[] generation (DashScope video-synthesis).
+// Sources (read 2026-09-29): alibabacloud.com/help/en/model-studio/
+// wan3-video-generation-api-reference, happyhorse-{text-to-video,image-to-video,
+// reference-to-video,video-edit}-api-reference, model-pricing (International).
+// Our resolution values stay lowercase; the service upper-cases them ("1080P").
+// ───────────────────────────────────────────────────────────────────────
+/** Durations 2–30s (Wan 3.0; with video input, input + output ≤ 30s). */
+const WAN3_DURATIONS = Array.from({ length: 29 }, (_, i) => i + 2);
+/** Durations 3–15s (HappyHorse 1.1). */
+const HAPPYHORSE_DURATIONS = Array.from({ length: 13 }, (_, i) => i + 3);
+const wan3Fields = {
+    prompt: {}, // prompt or media — the reference/frame modes can run on media alone
+    aspectRatio: { allowedValues: ["21:9", "16:9", "4:3", "1:1", "3:4", "9:16"] },
+    resolution: { allowedValues: ["480p", "720p", "1080p"], default: "1080p" },
+    duration: { allowedValues: WAN3_DURATIONS, default: 5 },
+    audio: {},
+    seed: {},
+    watermark: {},
+    promptOptimizer: {},
+    inputImageUrl: {}, // media first_frame
+    lastFrameImageUrl: {}, // media last_frame
+    referenceImageUrls: { max: 10 },
+    referenceVideoUrls: { max: 5 }, // ≤15s combined
+    referenceAudioUrls: { max: 5 }, // ≤15s combined
+    inputVideoUrl: {}, // sent as the first reference_video
+    // Document (`file`, ≤100MB / ≤50 pages) or public web page (`link`) the model
+    // parses into a video — one or the other, never with first/last frames.
+    // Forces prompt_extend on (API requirement).
+    referenceDocumentUrl: {},
+    referenceWebPageUrl: {},
+};
 exports.alibabaConfigs = {
+    // --- Wan 3.0 --- all-in-one (preview): text, first/last frame, or references.
+    "alibaba-wan-3.0": {
+        modelId: "wan3.0-video",
+        type: ["text-to-video", "image-to-video", "video-to-video"],
+        tier: "T3",
+        fields: wan3Fields,
+        cost: {
+            // International list price per second (a limited-time 30% discount is
+            // not applied here). Input reference-video seconds are billed too.
+            perSecondByResolution: { "480p": 0.05, "720p": 0.10, "1080p": 0.20 },
+            inputVideoPerSecond: { "480p": 0.05, "720p": 0.10, "1080p": 0.20 },
+        },
+        guidance: {
+            quality: 5,
+            speed: 2,
+            bestFor: "Long single shots (up to 30s, 30fps) and multimodal reference video: up to 10 images, 5 videos and 5 audio clips referenced in the prompt as \"Image 1\" / \"Video 1\" / \"Audio 1\"; also first/last frame.",
+            notes: "Preview. First/last frame can't be combined with references. $1.00 for 5s at 1080p, $0.50 at 720p. Generates audio by default.",
+        },
+    },
+    // --- Wan 3.0 Prime --- the higher-quality Wan 3.0 tier, same inputs.
+    "alibaba-wan-3.0-prime": {
+        modelId: "wan3.0-video-prime",
+        type: ["text-to-video", "image-to-video", "video-to-video"],
+        tier: "T3",
+        fields: wan3Fields,
+        cost: {
+            perSecondByResolution: { "480p": 0.068, "720p": 0.14, "1080p": 0.28 },
+            inputVideoPerSecond: { "480p": 0.068, "720p": 0.14, "1080p": 0.28 },
+        },
+        guidance: {
+            quality: 5,
+            speed: 1,
+            bestFor: "Hero shots where Wan 3.0 quality matters more than cost; same modes and reference limits as Wan 3.0.",
+            notes: "Preview. About 40% more than Wan 3.0: $1.40 for 5s at 1080p.",
+        },
+    },
+    // --- HappyHorse 1.1 t2v ---
+    "alibaba-happyhorse-1.1-t2v": {
+        modelId: "happyhorse-1.1-t2v",
+        type: ["text-to-video"],
+        tier: "T3",
+        fields: {
+            prompt: { required: true }, // ≤5000 chars
+            aspectRatio: { allowedValues: ["16:9", "9:16", "1:1", "4:3", "3:4", "4:5", "5:4", "9:21", "21:9"] },
+            resolution: { allowedValues: ["480p", "720p", "1080p"], default: "1080p" },
+            duration: { allowedValues: HAPPYHORSE_DURATIONS, default: 5 },
+            seed: {},
+            watermark: {},
+        },
+        cost: { perSecondByResolution: { "480p": 0.07, "720p": 0.14, "1080p": 0.18 } },
+        guidance: {
+            quality: 4,
+            speed: 3,
+            bestFor: "Cinematic text-to-video with the widest aspect set (incl. 4:5 / 5:4 social crops and 9:21); prompts in any language.",
+            notes: "$0.90 for 5s at 1080p, $0.35 at 480p. No first/last frame.",
+        },
+    },
+    // --- HappyHorse 1.1 i2v --- first frame only; output aspect follows the image.
+    "alibaba-happyhorse-1.1-i2v": {
+        modelId: "happyhorse-1.1-i2v",
+        type: ["image-to-video"],
+        tier: "T2",
+        acceptedImageFormats: ["image/jpeg", "image/jpg", "image/png", "image/webp", "image/bmp"],
+        fields: {
+            prompt: {},
+            inputImageUrl: { required: true }, // ≥300px, aspect 1:2.5–2.5:1, ≤20MB
+            resolution: { allowedValues: ["480p", "720p", "1080p"], default: "1080p" },
+            duration: { allowedValues: HAPPYHORSE_DURATIONS, default: 5 },
+            seed: {},
+            watermark: {},
+        },
+        cost: { perSecondByResolution: { "480p": 0.07, "720p": 0.14, "1080p": 0.18 } },
+        guidance: {
+            quality: 4,
+            speed: 3,
+            bestFor: "Animating a still from its first frame; prompt is optional.",
+            notes: "No last frame and no ratio (follows the image). $0.90 for 5s at 1080p.",
+        },
+    },
+    // --- HappyHorse 1.1 r2v --- 1–9 reference images, cited as "[Image 1]" etc.
+    "alibaba-happyhorse-1.1-r2v": {
+        modelId: "happyhorse-1.1-r2v",
+        type: ["text-to-video"],
+        tier: "T3",
+        fields: {
+            prompt: { required: true },
+            referenceImageUrls: { required: true, max: 9 }, // shortest side ≥400px
+            aspectRatio: { allowedValues: ["16:9", "9:16", "1:1", "4:3", "3:4", "4:5", "5:4", "9:21", "21:9"] },
+            resolution: { allowedValues: ["480p", "720p", "1080p"], default: "1080p" },
+            duration: { allowedValues: HAPPYHORSE_DURATIONS, default: 5 },
+            seed: {},
+            watermark: {},
+        },
+        cost: { perSecondByResolution: { "480p": 0.07, "720p": 0.14, "1080p": 0.18 } },
+        guidance: {
+            quality: 4,
+            speed: 3,
+            bestFor: "Keeping characters, products or styles consistent from up to 9 reference images, cited in the prompt as \"[Image 1]\", \"[Image 2]\"…",
+            notes: "Needs at least one reference image. $0.90 for 5s at 1080p.",
+        },
+    },
+    // --- HappyHorse 1.0 t2v / i2v / r2v --- the previous generation. Same
+    // request shape as 1.1 but no 480P tier and a pricier 1080P ($0.24/s vs
+    // $0.18/s), so 1.1 is the better default; kept for parity / A-B comparisons.
+    // EOL: none announced (checked 2026-09-29).
+    "alibaba-happyhorse-1.0-t2v": {
+        modelId: "happyhorse-1.0-t2v",
+        type: ["text-to-video"],
+        tier: "T3",
+        fields: {
+            prompt: { required: true }, // ≤5000 chars
+            aspectRatio: { allowedValues: ["16:9", "9:16", "1:1", "4:3", "3:4", "4:5", "5:4", "9:21", "21:9"] },
+            resolution: { allowedValues: ["720p", "1080p"], default: "1080p" },
+            duration: { allowedValues: HAPPYHORSE_DURATIONS, default: 5 },
+            seed: {},
+            watermark: {},
+        },
+        cost: { perSecondByResolution: { "720p": 0.14, "1080p": 0.24 } },
+        guidance: {
+            quality: 4,
+            speed: 3,
+            bestFor: "Text-to-video on the original HappyHorse model; same aspect set as 1.1.",
+            notes: "Previous generation: no 480p, and 1080p costs more than 1.1 ($1.20 vs $0.90 for 5s). Prefer 1.1 unless comparing.",
+        },
+    },
+    "alibaba-happyhorse-1.0-i2v": {
+        modelId: "happyhorse-1.0-i2v",
+        type: ["image-to-video"],
+        tier: "T2",
+        acceptedImageFormats: ["image/jpeg", "image/jpg", "image/png", "image/webp", "image/bmp"],
+        fields: {
+            prompt: {},
+            inputImageUrl: { required: true }, // ≥300px, aspect 1:2.5–2.5:1, ≤20MB
+            resolution: { allowedValues: ["720p", "1080p"], default: "1080p" },
+            duration: { allowedValues: HAPPYHORSE_DURATIONS, default: 5 },
+            seed: {},
+            watermark: {},
+        },
+        cost: { perSecondByResolution: { "720p": 0.14, "1080p": 0.24 } },
+        guidance: {
+            quality: 4,
+            speed: 3,
+            bestFor: "Animating a still from its first frame on the original HappyHorse model.",
+            notes: "Previous generation: no 480p, no last frame, ratio follows the image. $1.20 for 5s at 1080p; prefer 1.1.",
+        },
+    },
+    "alibaba-happyhorse-1.0-r2v": {
+        modelId: "happyhorse-1.0-r2v",
+        type: ["text-to-video"],
+        tier: "T3",
+        fields: {
+            prompt: { required: true },
+            referenceImageUrls: { required: true, max: 9 }, // shortest side ≥400px
+            aspectRatio: { allowedValues: ["16:9", "9:16", "1:1", "4:3", "3:4", "4:5", "5:4", "9:21", "21:9"] },
+            resolution: { allowedValues: ["720p", "1080p"], default: "1080p" },
+            duration: { allowedValues: HAPPYHORSE_DURATIONS, default: 5 },
+            seed: {},
+            watermark: {},
+        },
+        cost: { perSecondByResolution: { "720p": 0.14, "1080p": 0.24 } },
+        guidance: {
+            quality: 4,
+            speed: 3,
+            bestFor: "Consistent characters/products from up to 9 reference images (\"[Image 1]\"…) on the original HappyHorse model.",
+            notes: "Previous generation: no 480p. $1.20 for 5s at 1080p; prefer 1.1.",
+        },
+    },
+    // --- HappyHorse 1.0 video edit --- restyle/replace in an existing clip.
+    "alibaba-happyhorse-1.0-video-edit": {
+        modelId: "happyhorse-1.0-video-edit",
+        type: ["video-to-video"],
+        tier: "T3",
+        fields: {
+            prompt: { required: true },
+            inputVideoUrl: { required: true }, // MP4/MOV, 3–60s; only the first 15s is used
+            referenceImageUrls: { max: 5 },
+            resolution: { allowedValues: ["720p", "1080p"], default: "1080p" },
+            audio: {}, // false keeps the source audio (audio_setting "origin")
+            seed: {},
+            watermark: {},
+        },
+        cost: {
+            // List price; billed on input + output seconds. Output length matches
+            // the input (capped at 15s), so `duration` is the clip length.
+            perSecondByResolution: { "720p": 0.14, "1080p": 0.24 },
+            inputVideoPerSecond: { "720p": 0.14, "1080p": 0.24 },
+        },
+        guidance: {
+            quality: 4,
+            speed: 2,
+            bestFor: "Editing an existing clip by instruction (swap outfit, restyle, replace an object), optionally guided by up to 5 reference images.",
+            notes: "Output length follows the input (3–15s). Billed on input and output seconds: a 5s edit at 1080p is about $2.40.",
+        },
+    },
     // --- Wan 2.7 t2v --- newest text-to-video (Singapore region)
     "alibaba-wan-2.7-t2v": {
         modelId: "wan2.7-t2v",
@@ -20,34 +248,135 @@ exports.alibabaConfigs = {
         fields: {
             ...wanCommonFields,
             resolution: { allowedValues: ["720p", "1080p"] },
-            duration: { allowedValues: [5, 10] },
+            // duration: { allowedValues: [5, 10] }, // REMOVED FOR v2 — docs allow 2–15
+            duration: { allowedValues: WAN_2_15_DURATIONS },
+            inputAudioUrl: {}, // input.audio_url — WAV/MP3, 2–30s, drives the soundtrack
         },
-        cost: {
-            // Alibaba intl: 720p $0.10/s, 1080p $0.15/s.
-            table: {
-                "1280x720": { 5: 0.50, 10: 1.00 },
-                "1920x1080": { 5: 0.75, 10: 1.50 },
-            },
+        // REMOVED FOR v2 — kept for reference: a 5/10s table keyed by pixel size,
+        // which couldn't price the 2–15s range these models accept.
+        // cost: {
+        //   table: {
+        //     "1280x720":  { 5: 0.50, 10: 1.00 },
+        //     "1920x1080": { 5: 0.75, 10: 1.50 },
+        //   },
+        // },
+        // Alibaba intl: 720p $0.10/s, 1080p $0.15/s.
+        cost: { perSecondByResolution: { "720p": 0.10, "1080p": 0.15 } },
+        guidance: {
+            quality: 4,
+            speed: 3,
+            bestFor: "Text-to-video with generated audio, or a soundtrack driven by your own audio clip.",
+            notes: "Uses resolution + ratio (not a pixel size). $0.75 for 5s at 1080p.",
         },
     },
     // --- Wan 2.7 i2v --- newest image-to-video; supports first-last-frame interpolation
     "alibaba-wan-2.7-i2v": {
         modelId: "wan2.7-i2v",
-        type: ["image-to-video"],
+        // REMOVED FOR v2 — kept for reference: type: ["image-to-video"],
+        type: ["image-to-video", "video-to-video"], // video-to-video = first_clip continuation
         tier: "T3",
         fields: {
             ...wanCommonFields,
+            // No ratio: output follows the first frame / first clip.
+            aspectRatio: {},
             resolution: { allowedValues: ["720p", "1080p"] },
-            duration: { allowedValues: [5, 10] },
-            inputImageUrl: { required: true },
+            // duration: { allowedValues: [5, 10] }, // REMOVED FOR v2 — docs allow 2–15
+            duration: { allowedValues: WAN_2_15_DURATIONS },
+            // inputImageUrl: { required: true }, // REMOVED FOR v2 — a first clip can replace it
+            inputImageUrl: {}, // media first_frame
             lastFrameImageUrl: {}, // FLF interpolation
+            inputVideoUrl: {}, // media first_clip (2–10s) — continued up to `duration` total
+            inputAudioUrl: {}, // media driving_audio (WAV/MP3, 2–30s)
+        },
+        // REMOVED FOR v2 — kept for reference (5/10s table only):
+        // cost: {
+        //   table: {
+        //     "1280x720":  { 5: 0.50, 10: 1.00 },
+        //     "1920x1080": { 5: 0.75, 10: 1.50 },
+        //   },
+        // },
+        // Alibaba intl (with audio): 720p $0.10/s, 1080p $0.15/s. A first-clip
+        // continuation is billed on the full output `duration` (clip included).
+        cost: { perSecondByResolution: { "720p": 0.10, "1080p": 0.15 } },
+        guidance: {
+            quality: 4,
+            speed: 3,
+            bestFor: "Image-to-video with first+last frame, audio-driven motion (driving audio), or continuing an existing 2–10s clip.",
+            notes: "Output aspect follows the input. $0.75 for 5s at 1080p.",
+        },
+    },
+    // --- Wan 2.7 r2v --- multi-subject reference-to-video with per-subject voices.
+    // Source: alibabacloud.com/help/en/model-studio/wan-video-to-video-api-reference
+    // (read 2026-09-29). input.media[]: ≤1 first_frame + 1–5 reference images/videos
+    // (combined), each image/video optionally carrying a `reference_voice`.
+    // EOL: none announced (checked 2026-09-29).
+    "alibaba-wan-2.7-r2v": {
+        modelId: "wan2.7-r2v",
+        type: ["text-to-video", "image-to-video"],
+        tier: "T3",
+        fields: {
+            prompt: { required: true }, // ≤5000 chars; cite "Image 1" / "Video 1"
+            negative_prompt: {},
+            aspectRatio: { allowedValues: ["16:9", "9:16", "1:1", "4:3", "3:4"] }, // ignored with a first frame
+            resolution: { allowedValues: ["720p", "1080p"], default: "1080p" },
+            duration: { allowedValues: WAN_2_15_DURATIONS, default: 5 }, // 2–10 when a reference video is present
+            inputImageUrl: {}, // media first_frame (optional composition control)
+            referenceImageUrls: { max: 5 }, // images + videos ≤ 5 combined
+            referenceVideoUrls: { max: 5 }, // MP4/MOV 1–30s; the subject's voice is reused
+            // Voices (WAV/MP3, 1–10s), matched in order to the subjects as sent:
+            // reference videos first, then reference images. Overrides a video's audio.
+            referenceAudioUrls: { max: 5 },
+            seed: {},
+            watermark: {},
+            promptOptimizer: {},
+        },
+        cost: {
+            // Input and output seconds both billed (International list). Input video
+            // is capped at 5s total (5s ÷ number of reference videos, per video) —
+            // min(total, 5) is the safe upper bound we charge.
+            perSecondByResolution: { "720p": 0.10, "1080p": 0.15 },
+            inputVideoPerSecond: { "720p": 0.10, "1080p": 0.15 },
+            inputVideoBillableMaxSec: 5,
+        },
+        guidance: {
+            quality: 4,
+            speed: 3,
+            bestFor: "Multi-character scenes from up to 5 reference images/videos (cited \"Image 1\" / \"Video 1\"), each with its own voice; accepts a storyboard grid image; optional first frame.",
+            notes: "Images are free; reference videos add at most 5 billed seconds. $0.75 for 5s at 1080p. Max 10s when a reference video is used.",
+        },
+    },
+    // --- Wan 2.7 video edit --- instruction edit of a 2–10s clip.
+    // Source: alibabacloud.com/help/en/model-studio/wan-video-editing-api-reference
+    // (read 2026-09-29). EOL: none announced (checked 2026-09-29).
+    "alibaba-wan-2.7-videoedit": {
+        modelId: "wan2.7-videoedit",
+        type: ["video-to-video"],
+        tier: "T3",
+        fields: {
+            prompt: {}, // the edit instruction (≤5000 chars)
+            negative_prompt: {},
+            inputVideoUrl: { required: true }, // MP4/MOV, 2–10s, ≤100MB
+            referenceImageUrls: { max: 4 },
+            // Omitted → keeps the input's shape; set → forces this ratio.
+            aspectRatio: { allowedValues: ["16:9", "9:16", "1:1", "4:3", "3:4"] },
+            resolution: { allowedValues: ["720p", "1080p"], default: "1080p" },
+            // Output = the input clip; a shorter duration truncates it from 0s.
+            duration: { allowedValues: Array.from({ length: 9 }, (_, i) => i + 2), default: 5 },
+            audio: {}, // false keeps the source audio (audio_setting "origin")
+            seed: {},
+            watermark: {},
+            promptOptimizer: {},
         },
         cost: {
-            // Alibaba intl (with audio): 720p $0.10/s, 1080p $0.15/s.
-            table: {
-                "1280x720": { 5: 0.50, 10: 1.00 },
-                "1920x1080": { 5: 0.75, 10: 1.50 },
-            },
+            // Billed on input + output seconds (International list).
+            perSecondByResolution: { "720p": 0.10, "1080p": 0.15 },
+            inputVideoPerSecond: { "720p": 0.10, "1080p": 0.15 },
+        },
+        guidance: {
+            quality: 4,
+            speed: 3,
+            bestFor: "Cheaper instruction edits of a short clip (restyle, swap clothes, add an object from up to 4 reference images).",
+            notes: "Clip must be 2–10s. Billed on input + output: a 5s edit at 1080p is $1.50 (vs about $2.40 on HappyHorse edit).",
         },
     },
     // --- Wan 2.6 t2v --- 720p/1080p only (no 480p); 5/10s
@@ -58,15 +387,20 @@ exports.alibabaConfigs = {
         fields: {
             ...wanCommonFields,
             resolution: { allowedValues: ["720p", "1080p"] },
-            duration: { allowedValues: [5, 10] },
-        },
-        cost: {
-            // Alibaba intl: 720p $0.10/s, 1080p $0.15/s.
-            table: {
-                "1280x720": { 5: 0.50, 10: 1.00 },
-                "1920x1080": { 5: 0.75, 10: 1.50 },
-            },
+            // duration: { allowedValues: [5, 10] }, // REMOVED FOR v2 — docs allow 2–15
+            duration: { allowedValues: WAN_2_15_DURATIONS },
+            inputAudioUrl: {}, // input.audio_url
         },
+        // REMOVED FOR v2 — kept for reference: a 5/10s table keyed by pixel size,
+        // which couldn't price the 2–15s range these models accept.
+        // cost: {
+        //   table: {
+        //     "1280x720":  { 5: 0.50, 10: 1.00 },
+        //     "1920x1080": { 5: 0.75, 10: 1.50 },
+        //   },
+        // },
+        // Alibaba intl: 720p $0.10/s, 1080p $0.15/s.
+        cost: { perSecondByResolution: { "720p": 0.10, "1080p": 0.15 } },
     },
     // --- Wan 2.6 i2v --- image-to-video with audio
     "alibaba-wan-2.6-i2v": {
@@ -75,17 +409,23 @@ exports.alibabaConfigs = {
         tier: "T3",
         fields: {
             ...wanCommonFields,
+            aspectRatio: {}, // output follows the image
             resolution: { allowedValues: ["720p", "1080p"] },
-            duration: { allowedValues: [5, 10] },
+            // duration: { allowedValues: [5, 10] }, // REMOVED FOR v2 — docs allow 2–15
+            duration: { allowedValues: WAN_2_15_DURATIONS },
             inputImageUrl: { required: true },
+            inputAudioUrl: {}, // input.audio_url
         },
-        cost: {
-            // Alibaba intl: 720p $0.10/s, 1080p $0.15/s.
-            table: {
-                "1280x720": { 5: 0.50, 10: 1.00 },
-                "1920x1080": { 5: 0.75, 10: 1.50 },
-            },
-        },
+        // REMOVED FOR v2 — kept for reference: a 5/10s table keyed by pixel size,
+        // which couldn't price the 2–15s range these models accept.
+        // cost: {
+        //   table: {
+        //     "1280x720":  { 5: 0.50, 10: 1.00 },
+        //     "1920x1080": { 5: 0.75, 10: 1.50 },
+        //   },
+        // },
+        // Alibaba intl: 720p $0.10/s, 1080p $0.15/s.
+        cost: { perSecondByResolution: { "720p": 0.10, "1080p": 0.15 } },
     },
     // --- Wan 2.6 i2v flash --- ~50% cheaper i2v variant (audio optional)
     "alibaba-wan-2.6-i2v-flash": {
@@ -94,16 +434,25 @@ exports.alibabaConfigs = {
         tier: "T2",
         fields: {
             ...wanCommonFields,
+            aspectRatio: {}, // output follows the image
             resolution: { allowedValues: ["720p", "1080p"] },
-            duration: { allowedValues: [5, 10] },
+            // duration: { allowedValues: [5, 10] }, // REMOVED FOR v2 — docs allow 2–15
+            duration: { allowedValues: WAN_2_15_DURATIONS },
             inputImageUrl: { required: true },
+            inputAudioUrl: {}, // input.audio_url
+            audio: {}, // false = silent video at half price
         },
+        // REMOVED FOR v2 — kept for reference (5/10s table, audio price only):
+        // cost: {
+        //   table: {
+        //     "1280x720":  { 5: 0.25, 10: 0.50 },
+        //     "1920x1080": { 5: 0.38, 10: 0.75 },
+        //   },
+        // },
+        // Alibaba intl: audio 720p $0.05/s, 1080p $0.075/s; silent is half that.
         cost: {
-            // Alibaba intl (with audio): 720p $0.05/s, 1080p $0.075/s.
-            table: {
-                "1280x720": { 5: 0.25, 10: 0.50 },
-                "1920x1080": { 5: 0.38, 10: 0.75 },
-            },
+            perSecondByResolution: { "720p": 0.05, "1080p": 0.075 },
+            silentPerSecondByResolution: { "720p": 0.025, "1080p": 0.0375 },
         },
     },
     // --- Wan 2.5 t2v --- preview tier
@@ -111,15 +460,19 @@ exports.alibabaConfigs = {
         modelId: "wan2.5-t2v-preview",
         type: ["text-to-video"],
         tier: "T2",
-        fields: { ...wanCommonFields },
-        cost: {
-            // Alibaba intl: 480p $0.05/s, 720p $0.10/s, 1080p $0.15/s.
-            table: {
-                "832x480": { 5: 0.25, 10: 0.50 },
-                "1280x720": { 5: 0.50, 10: 1.00 },
-                "1920x1080": { 5: 0.75, 10: 1.50 },
-            },
-        },
+        fields: { ...wanCommonFields, inputAudioUrl: {} }, // input.audio_url
+        // REMOVED FOR v2 — kept for reference. Keyed "832x480" while
+        // getAlibabaDimensions returns "832*480" (since 2026-08-16), so every
+        // lookup missed and billed the $1.00 fallback.
+        // cost: {
+        //   table: {
+        //     "832x480":   { 5: 0.25, 10: 0.50 },
+        //     "1280x720":  { 5: 0.50, 10: 1.00 },
+        //     "1920x1080": { 5: 0.75, 10: 1.50 },
+        //   },
+        // },
+        // Alibaba intl: 480p $0.05/s, 720p $0.10/s, 1080p $0.15/s.
+        cost: { perSecondByResolution: { "480p": 0.05, "720p": 0.10, "1080p": 0.15 } },
     },
     // --- Wan 2.2 t2v --- 480p + 1080p only (no 720p tier officially priced)
     "alibaba-wan-2.2-t2v": {
@@ -129,14 +482,18 @@ exports.alibabaConfigs = {
         fields: {
             ...wanCommonFields,
             resolution: { allowedValues: ["480p", "1080p"] },
+            duration: { allowedValues: [5] }, // fixed at 5s per the docs; 10 was rejected
         },
-        cost: {
-            // Alibaba intl (matches i2v-plus tier): 480p $0.02/s, 1080p $0.10/s.
-            table: {
-                "832x480": { 5: 0.10, 10: 0.20 },
-                "1920x1080": { 5: 0.50, 10: 1.00 },
-            },
-        },
+        // REMOVED FOR v2 — kept for reference. Same "x" vs "*" key mismatch as
+        // wan 2.5 (always billed the fallback), and 10s isn't offered.
+        // cost: {
+        //   table: {
+        //     "832x480":   { 5: 0.10, 10: 0.20 },
+        //     "1920x1080": { 5: 0.50, 10: 1.00 },
+        //   },
+        // },
+        // Alibaba intl: 480p $0.02/s, 1080p $0.10/s.
+        cost: { perSecondByResolution: { "480p": 0.02, "1080p": 0.10 } },
     },
     // --- Wan 2.2 s2v --- REMOVED FOR v1 — China-region only, intl key cannot reach
     //
@@ -177,6 +534,56 @@ exports.alibabaConfigs = {
     // ───────────────────────────────────────────────────────────────────────
     // Image-gen — Qwen-Image + Kolors (both via DashScope text2image / image2image)
     // ───────────────────────────────────────────────────────────────────────
+    // --- Qwen-Image 3.0 --- one model id for text-to-image AND editing (1–3
+    // input images), up to 6 outputs, strong layout/small-text rendering.
+    // Sync multimodal endpoint, same message format as 2.0. Priced by output
+    // tier (our aspect presets are 2k-class) plus $0.003 per input image.
+    // Source: qwen-image-generation-and-editing-api-reference + model-pricing,
+    // read 2026-09-29.
+    "alibaba-qwen-image-3.0-pro": {
+        modelId: "qwen-image-3.0-pro",
+        type: ["text-to-image", "image-to-image"],
+        tier: "T2",
+        fields: {
+            prompt: { required: true },
+            aspectRatio: { allowedValues: ["1:1", "9:16", "16:9", "3:4", "4:3", "2:3", "3:2"] },
+            numImages: { default: 1 }, // 1–6
+            inputImageUrls: { max: 3 },
+            seed: {},
+            watermark: {},
+            promptOptimizer: {},
+            negativePrompt: {},
+        },
+        cost: { fixed: 0.075, perInputImage: 0.003 }, // 2k; 1k is $0.04
+        guidance: {
+            quality: 5,
+            speed: 3,
+            bestFor: "Dense layouts and legible small text (posters, menus, storyboards, infographics) in many languages; also multi-image edits.",
+            notes: "$0.075 per 2k image. Up to 6 images per call.",
+        },
+    },
+    "alibaba-qwen-image-3.0": {
+        modelId: "qwen-image-3.0",
+        type: ["text-to-image", "image-to-image"],
+        tier: "T1",
+        fields: {
+            prompt: { required: true },
+            aspectRatio: { allowedValues: ["1:1", "9:16", "16:9", "3:4", "4:3", "2:3", "3:2"] },
+            numImages: { default: 1 },
+            inputImageUrls: { max: 3 },
+            seed: {},
+            watermark: {},
+            promptOptimizer: {},
+            negativePrompt: {},
+        },
+        cost: { fixed: 0.03, perInputImage: 0.003 }, // same price at 1k and 2k
+        guidance: {
+            quality: 4,
+            speed: 4,
+            bestFor: "Faster, cheaper Qwen-Image 3.0 for text-heavy images and edits.",
+            notes: "$0.03 per image at 1k or 2k.",
+        },
+    },
     "alibaba-qwen-image-2.0-pro": {
         modelId: "qwen-image-2.0-pro",
         type: ["text-to-image"],

package/lib/globals/aiModels/providers/bytedance.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"bytedance.d.ts","sourceRoot":"","sources":["../../../../src/globals/aiModels/providers/bytedance.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,WAAW,EAAE,MAAM,UAAU,CAAC;AACvC,OAAO,EAAE,YAAY,EAAE,MAAM,UAAU,CAAC;AAExC,eAAO,MAAM,gBAAgB,EAAE,OAAO,CAAC,MAAM,CAAC,WAAW,EAAE,YAAY,CAAC,CA4OvE,CAAC"}
+{"version":3,"file":"bytedance.d.ts","sourceRoot":"","sources":["../../../../src/globals/aiModels/providers/bytedance.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,WAAW,EAAE,MAAM,UAAU,CAAC;AACvC,OAAO,EAAE,YAAY,EAAE,MAAM,UAAU,CAAC;AAWxC,eAAO,MAAM,gBAAgB,EAAE,OAAO,CAAC,MAAM,CAAC,WAAW,EAAE,YAAY,CAAC,CA+ZvE,CAAC"}