vidspotai-shared 1.0.67 → 1.0.69-dev.0

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

@@ -5,31 +5,57 @@ exports.googleConfigs = {
     "google-veo-3.1": {
         modelId: "veo-3.1-generate-preview",
         type: ["text-to-video", "image-to-video"],
+        tier: "T3",
         fields: {
             prompt: { required: true },
             aspectRatio: { allowedValues: ["16:9", "9:16"] },
-            resolution: { allowedValues: ["720p", "1080p"] },
+            // 4k requires durationSeconds=8 and is incompatible with referenceImages.
+            resolution: { allowedValues: ["720p", "1080p", "4k"] },
             duration: { allowedValues: [4, 6, 8] },
             imageUrl: {},
+            lastFrameImageUrl: {},
+            // Up to 3 reference images for character/scene consistency. Forces duration=8.
+            referenceImageUrls: { max: 3 },
+            // Video extension — Gemini API restriction: input MUST be a Veo-generated
+            // Video reference from a prior generate_videos operation (not an arbitrary
+            // uploaded MP4 / GCS URI / videoBytes). Source TTL is 2 days, resets on
+            // each extend. Output capped at 720p, +~7s per call, max cumulative 141s.
+            inputVideoUrl: {},
             negative_prompt: {},
             audio: {},
+            seed: {},
+            personGeneration: { allowedValues: ["allow_all", "allow_adult", "dont_allow"] },
+            promptOptimizer: {},
         },
         requestPerMin: 50,
         cost: {
-            perSecond: 0.4,
+            // Vertex: $0.40/s @ 720p/1080p, $0.60/s @ 4k.
+            perResolution: { "720p": 0.40, "1080p": 0.40, "4k": 0.60 },
         },
     },
     "google-veo-3.1-fast": {
         modelId: "veo-3.1-fast-generate-preview",
         type: ["text-to-video", "image-to-video"],
+        tier: "T3",
         fields: {
             prompt: { required: true },
             aspectRatio: { allowedValues: ["16:9", "9:16"] },
             resolution: { allowedValues: ["720p", "1080p"] },
             duration: { allowedValues: [4, 6, 8] },
             imageUrl: {},
+            lastFrameImageUrl: {},
+            // Up to 3 reference images for character/scene consistency. Forces duration=8.
+            referenceImageUrls: { max: 3 },
+            // Video extension — Gemini API restriction: input MUST be a Veo-generated
+            // Video reference from a prior generate_videos operation (not an arbitrary
+            // uploaded MP4 / GCS URI / videoBytes). Source TTL 2 days, resets on each
+            // extend. Output capped at 720p, +~7s per call, max cumulative 141s.
+            inputVideoUrl: {},
             negative_prompt: {},
             audio: {},
+            seed: {},
+            personGeneration: { allowedValues: ["allow_all", "allow_adult", "dont_allow"] },
+            promptOptimizer: {},
         },
         requestPerMin: 50,
         cost: {
@@ -39,14 +65,21 @@ exports.googleConfigs = {
     "google-veo-3.1-lite": {
         modelId: "veo-3.1-lite-generate-preview",
         type: ["text-to-video", "image-to-video"],
+        tier: "T2",
         fields: {
             prompt: { required: true },
             aspectRatio: { allowedValues: ["16:9", "9:16"] },
             resolution: { allowedValues: ["720p", "1080p"] },
             duration: { allowedValues: [4, 6, 8] },
             imageUrl: {},
+            lastFrameImageUrl: {},
+            // Lite does NOT support reference images or video extension per the
+            // Gemini API spec — only 3.1 and 3.1-fast do.
             negative_prompt: {},
             audio: {},
+            seed: {},
+            personGeneration: { allowedValues: ["allow_all", "allow_adult", "dont_allow"] },
+            promptOptimizer: {},
         },
         requestPerMin: 50,
         cost: {
@@ -109,4 +142,119 @@ exports.googleConfigs = {
     //   requestPerMin: 10,
     //   cost: { perSecond: 0.35 },
     // },
+    // ───────────────────────────────────────────────────────────────────────
+    // Image-gen — Imagen 4 (raw T2I) + Nano Banana (T2I + instruction edit)
+    // ───────────────────────────────────────────────────────────────────────
+    // Imagen 4 — text-to-image only. No image input. aspectRatio drives output dims.
+    "google-imagen-4-ultra": {
+        modelId: "imagen-4.0-ultra-generate-001",
+        type: ["text-to-image"],
+        tier: "T2",
+        fields: {
+            prompt: { required: true },
+            aspectRatio: { allowedValues: ["1:1", "9:16", "16:9", "3:4", "4:3"] },
+            numImages: { default: 1 },
+            personGeneration: { allowedValues: ["allow_all", "allow_adult", "dont_allow"] },
+        },
+        cost: { fixed: 0.06 }, // per image
+    },
+    "google-imagen-4": {
+        modelId: "imagen-4.0-generate-001",
+        type: ["text-to-image"],
+        tier: "T2",
+        fields: {
+            prompt: { required: true },
+            aspectRatio: { allowedValues: ["1:1", "9:16", "16:9", "3:4", "4:3"] },
+            numImages: { default: 1 },
+            personGeneration: { allowedValues: ["allow_all", "allow_adult", "dont_allow"] },
+        },
+        cost: { fixed: 0.04 },
+    },
+    "google-imagen-4-fast": {
+        modelId: "imagen-4.0-fast-generate-001",
+        type: ["text-to-image"],
+        tier: "T1",
+        fields: {
+            prompt: { required: true },
+            aspectRatio: { allowedValues: ["1:1", "9:16", "16:9", "3:4", "4:3"] },
+            numImages: { default: 1 },
+            personGeneration: { allowedValues: ["allow_all", "allow_adult", "dont_allow"] },
+        },
+        cost: { fixed: 0.02 },
+    },
+    // Nano Banana — Gemini image models. Support T2I and instruction-edit
+    // (input images supplied via inputImageUrls; up to 14 refs for Banana 2/Pro).
+    "google-nano-banana-pro": {
+        modelId: "gemini-3-pro-image-preview",
+        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"] },
+            inputImageUrls: { max: 14 },
+            imageSize: { allowedValues: ["1K", "2K", "4K", "auto_2K"] },
+        },
+        // Vertex (gemini-3-pro-image-preview): output charged per-image by resolution.
+        // 1K & 2K (1MP/4MP) = 1120 tokens → $0.134; 4K (16MP) = 2000 tokens → $0.24.
+        cost: {
+            perImageSize: { "1K": 0.134, "2K": 0.134, "auto_2K": 0.134, "4K": 0.24 },
+            fixed: 0.134, // fallback when imageSize not supplied
+        },
+    },
+    // (Removed `google-nano-banana-2` — `gemini-2.5-pro-image` does not exist on
+    // Vertex. The 2.5 family only ships `gemini-2.5-flash-image` which is the
+    // current `google-nano-banana` entry below.)
+    "google-nano-banana": {
+        modelId: "gemini-2.5-flash-image",
+        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"] },
+            inputImageUrls: {},
+        },
+        cost: { fixed: 0.0387 }, // ~$0.04 per image; per Google pricing (1290 tokens × $30/1M)
+    },
+    // ───────────────────────────────────────────────────────────────────────
+    // Music-gen — Lyria 2 (Vertex AI predict endpoint, ADC auth)
+    // ───────────────────────────────────────────────────────────────────────
+    "google-lyria-2": {
+        modelId: "lyria-002",
+        type: ["text-to-music"],
+        tier: "T1",
+        fields: {
+            prompt: { required: true },
+            negativePrompt: {},
+            seed: {},
+            numImages: { default: 1 }, // sample_count — number of music clips returned
+        },
+        // Vertex Lyria 2 — flat per-clip pricing (lyria-002 still live, not retired).
+        cost: { fixed: 0.06 },
+    },
+    // Lyria 3 — newer / cheaper alternative.
+    "google-lyria-3": {
+        modelId: "lyria-003",
+        type: ["text-to-music"],
+        tier: "T1",
+        fields: {
+            prompt: { required: true },
+            negativePrompt: {},
+            seed: {},
+            numImages: { default: 1 },
+        },
+        cost: { fixed: 0.04 }, // per 30-sec clip
+    },
+    // Lyria 3 Pro — premium full-song tier (≤ 3 min).
+    "google-lyria-3-pro": {
+        modelId: "lyria-3-pro",
+        type: ["text-to-music"],
+        tier: "T2",
+        fields: {
+            prompt: { required: true },
+            negativePrompt: {},
+            seed: {},
+            numImages: { default: 1 },
+        },
+        cost: { fixed: 0.08 }, // per full song (up to 3 min)
+    },
 };

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

@@ -1 +1 @@
-{"version":3,"file":"kling.d.ts","sourceRoot":"","sources":["../../../../src/globals/aiModels/providers/kling.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,WAAW,EAAE,MAAM,UAAU,CAAC;AACvC,OAAO,EAAsB,YAAY,EAAE,MAAM,UAAU,CAAC;AAE5D,eAAO,MAAM,YAAY,EAAE,OAAO,CAAC,MAAM,CAAC,WAAW,EAAE,YAAY,CAAC,CA2InE,CAAC"}
+{"version":3,"file":"kling.d.ts","sourceRoot":"","sources":["../../../../src/globals/aiModels/providers/kling.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,WAAW,EAAE,MAAM,UAAU,CAAC;AACvC,OAAO,EAAsB,YAAY,EAAE,MAAM,UAAU,CAAC;AAqC5D,eAAO,MAAM,YAAY,EAAE,OAAO,CAAC,MAAM,CAAC,WAAW,EAAE,YAAY,CAAC,CAsWnE,CAAC"}

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

@@ -1,31 +1,274 @@
 "use strict";
 Object.defineProperty(exports, "__esModule", { value: true });
 exports.klingConfigs = void 0;
+// Fields shared across all standard Kling video models. I2V-capable models
+// also pick up image_tail / static_mask / dynamic_masks via spread below.
+const klingCommonFields = {
+    prompt: { required: true },
+    aspectRatio: { allowedValues: ["16:9", "9:16", "1:1"] },
+    duration: { allowedValues: [5, 10] },
+    negative_prompt: {},
+    cfgScale: {},
+    callbackUrl: {},
+    externalTaskId: {},
+};
+const klingI2vFields = {
+    inputImageUrl: {},
+    lastFrameImageUrl: {}, // image_tail
+    staticMaskUrl: {},
+    dynamicMasks: {},
+    cameraControl: {},
+};
+// ─── Kling Image Generation ───────────────────────────────────────────────────
+// Two separate endpoint families per official docs (2026-05-20):
+//   /v1/images/omni-image   → kling-image-o1, kling-v3-omni  (Omni-Image / IMAGE 3.0)
+//   /v1/images/generations  → kling-v2-1-image               (standard gen / IMAGE 2.1)
+// Auth is the same JWT as video (KLING_ACCESS_KEY + KLING_SECRET_KEY).
+// Pricing below is UNVERIFIED — set to safe-side estimate. Confirm from Kling dashboard.
+const klingImageCommonFields = {
+    prompt: { required: true },
+    callbackUrl: {},
+    externalTaskId: {},
+    numImages: { default: 1 }, // n: 1–9
+    watermark: {}, // watermark_info.enabled
+};
 exports.klingConfigs = {
-    // "kling-video-o1": {}
-    "kling-v2.6": {
-        modelId: "kling-v2-6",
+    // ─── IMAGE 3.0 (Omni-Image API: POST /v1/images/omni-image) ───────────────
+    // kling-image-o1 — Kling's newest image model; supports T2I, I2I,
+    // multi-image element control. Custom aspect ratio 1K/2K.
+    // Source: https://kling.ai/document-api/apiReference/model/OmniImage
+    "kling-image-o1": {
+        modelId: "kling-image-o1",
+        type: ["text-to-image", "image-to-image"],
+        tier: "T1",
+        fields: {
+            ...klingImageCommonFields,
+            // aspect_ratio: 16:9, 9:16, 1:1, 4:3, 3:4, 3:2, 2:3, 21:9, auto
+            aspectRatio: { allowedValues: ["16:9", "9:16", "1:1", "4:3", "3:4", "3:2", "2:3", "21:9", "auto"] },
+            resolution: { allowedValues: ["1k", "2k"] },
+            inputImageUrls: {}, // image_list[] — multi-image references (max 10 combined with elements)
+        },
+        concurrentRequests: 5,
+        // Verified 2026-05-20 from kling.ai/dev/pricing: 8 units × $0.0035/unit = $0.028 (1K and 2K).
+        cost: { fixed: 0.028 },
+    },
+    // kling-v3-omni — extends kling-image-o1 with 4K resolution + series generation.
+    // Source: https://kling.ai/document-api/apiReference/model/OmniImage
+    "kling-v3-omni": {
+        modelId: "kling-v3-omni",
+        type: ["text-to-image", "image-to-image"],
+        tier: "T1",
+        fields: {
+            ...klingImageCommonFields,
+            aspectRatio: { allowedValues: ["16:9", "9:16", "1:1", "4:3", "3:4", "3:2", "2:3", "21:9", "auto"] },
+            resolution: { allowedValues: ["1k", "2k", "4k"] },
+            inputImageUrls: {}, // image_list[]
+        },
+        concurrentRequests: 5,
+        // Verified 2026-05-20 from kling.ai/dev/pricing: 1K/2K = 8 units = $0.028; 4K = 16 units = $0.056.
+        // Using 1K/2K rate as base. getCreditUsed() uses this fixed value; 4K over-billing is acceptable
+        // until we add resolution-based cost routing.
+        cost: { fixed: 0.028 },
+    },
+    // ─── IMAGE 2.1 (standard gen API: POST /v1/images/generations) ───────────
+    // kling-v2-1-image — image key distinct from video "kling-v2.1".
+    // Source: https://kling.ai/document-api/apiReference/model/imageGeneration
+    "kling-v2-1-image": {
+        modelId: "kling-v2-1", // Kling API model_name value
+        type: ["text-to-image", "image-to-image"],
+        tier: "T1",
+        fields: {
+            ...klingImageCommonFields,
+            aspectRatio: { allowedValues: ["16:9", "9:16", "1:1", "4:3", "3:4", "3:2", "2:3", "21:9"] },
+            resolution: { allowedValues: ["1k", "2k"] },
+            negativePrompt: {}, // only valid for T2I (not I2I)
+            inputImageUrl: {}, // single reference image (Base64 or URL)
+            inputImageUrls: {}, // multi-image references (multi-image-to-image mode)
+            // image_reference: subject | face — used when inputImageUrl is set
+            imageReference: { allowedValues: ["subject", "face"] },
+        },
+        concurrentRequests: 5,
+        // Verified 2026-05-20 from kling.ai/dev/pricing: T2I = 4 units = $0.014; I2I = 8 units = $0.028.
+        // Using T2I rate as base; I2I tasks are charged at $0.028 by Kling but we credit at T2I rate.
+        cost: { fixed: 0.014 },
+    },
+    // ─── OmniVideo models (/v1/videos/omni-video) ────────────────────────────
+    // kling-video-o1 — Kling's reasoning video model.
+    // std/pro, 5s or 10s only. T2V + I2V (start frame + end frame) + element control + video reference.
+    // Docs: https://kling.ai/document-api/apiReference/model/OmniVideo
+    "kling-video-o1": {
+        modelId: "kling-video-o1",
         type: ["text-to-video", "image-to-video"],
+        tier: "T3",
         fields: {
-            prompt: { required: true },
+            prompt: {},
             aspectRatio: { allowedValues: ["16:9", "9:16", "1:1"] },
             duration: { allowedValues: [5, 10] },
             mode: { allowedValues: ["std", "pro"] },
             negative_prompt: {},
+            callbackUrl: {},
+            externalTaskId: {},
+            // OmniVideo uses image_list[] instead of the image/image_tail fields
+            referenceImageUrls: {}, // image_list entries without a frame type
+            inputImageUrl: {}, // image_list entry with type: first_frame
+            lastFrameImageUrl: {}, // image_list entry with type: end_frame
+            referenceVideoUrls: {}, // video_list[] (up to 1)
+            multiShot: {},
+        },
+        concurrentRequests: 5,
+        // Verified 2026-05-20 from kling.ai/dev/pricing (Video-O1 tab): std=$0.084/s (T2V), pro=$0.112/s.
+        // I2V (with video input) costs 50% more: std=$0.126/s, pro=$0.168/s (no separate table key yet).
+        cost: {
+            table: {
+                std: { 5: 0.42, 10: 0.84 },
+                pro: { 5: 0.56, 10: 1.12 },
+            },
+        },
+    },
+    // kling-v3-omni VIDEO — full OmniVideo with 4K + multi-shot + 3–15s.
+    // "kling-v3-omni" as IMAGE uses /v1/images/omni-image (KlingImageService).
+    // "kling-v3-omni-video" as VIDEO uses /v1/videos/omni-video (KlingService).
+    // Backend key is "kling-v3-omni-video"; Kling API model_name is "kling-v3-omni".
+    "kling-v3-omni-video": {
+        modelId: "kling-v3-omni",
+        type: ["text-to-video", "image-to-video"],
+        tier: "T3",
+        fields: {
+            prompt: {},
+            aspectRatio: { allowedValues: ["16:9", "9:16", "1:1"] },
+            duration: { allowedValues: [3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15] },
+            mode: { allowedValues: ["std", "pro", "4k"] },
+            negative_prompt: {},
+            callbackUrl: {},
+            externalTaskId: {},
+            referenceImageUrls: {},
+            inputImageUrl: {},
+            lastFrameImageUrl: {},
+            referenceVideoUrls: {},
+            multiShot: {},
+        },
+        concurrentRequests: 5,
+        // Pricing verified 2026-05-20 from V3-Omni tab:
+        //   std T2V no-audio: $0.084/s → 5s=$0.42, 10s=$0.84
+        //   pro T2V no-audio: $0.112/s → 5s=$0.56, 10s=$1.12
+        //   4k: $0.42/s → 5s=$2.10, 10s=$4.20
+        // Using no-audio T2V rates as base (I2V adds 50% more per Kling pricing table).
+        cost: {
+            table: {
+                std: { 5: 0.42, 10: 0.84 },
+                pro: { 5: 0.56, 10: 1.12 },
+                "4k": { 5: 2.10, 10: 4.20 },
+            },
+        },
+    },
+    // ─── Motion Control (/v1/videos/motion-control) ───────────────────────────
+    // Takes a character reference image + a motion reference video.
+    // Supported models: kling-v2-6, kling-v3. Uses kling-v3 by default (higher quality).
+    // Duration = reference video duration (3–10s when character_orientation="image",
+    // 3–30s when character_orientation="video").
+    "kling-motion-control": {
+        modelId: "kling-v3", // default model; can be overridden per request
+        type: ["motion-control"],
+        tier: "T2",
+        fields: {
+            inputImageUrl: { required: true }, // character reference image (image_url)
+            inputVideoUrl: { required: true }, // motion reference video (video_url)
+            prompt: {},
+            mode: { allowedValues: ["std", "pro"] },
+            callbackUrl: {},
+            externalTaskId: {},
+        },
+        concurrentRequests: 5,
+        // Verified 2026-05-20 from kling.ai/dev/pricing (V3 tab, Motion Control row):
+        // kling-v3 std=$0.126/s, pro=$0.168/s. Duration = reference video length (variable; 5s used as base).
+        cost: {
+            table: {
+                std: { 5: 0.63, 10: 1.26 },
+                pro: { 5: 0.84, 10: 1.68 },
+            },
+        },
+    },
+    // ─── Lip Sync (/v1/videos/identify-face + /v1/videos/advanced-lip-sync) ──
+    // Two-step: identify faces in source video → apply audio lip-sync.
+    // Input: source video URL + audio URL.
+    "kling-lip-sync": {
+        modelId: "kling-lip-sync",
+        type: ["lip-sync"],
+        tier: "T2",
+        fields: {
+            inputVideoUrl: { required: true }, // source video (passed to identify-face)
+            inputAudioUrl: { required: true }, // audio file (sound_file in advanced-lip-sync)
+            prompt: {},
+            callbackUrl: {},
+            externalTaskId: {},
+        },
+        concurrentRequests: 5,
+        // Verified 2026-05-20 from kling.ai/dev/pricing (Functional Models tab):
+        // $0.07 per 5s (rounded up to next 5s increment). Using fixed = one 5s block.
+        cost: { fixed: 0.07 },
+    },
+    // Kling v3.0 — adds `generate_audio` (boolean, native audio generation) and
+    // `multi_prompt[]` + `shot_type` for multi-shot videos. Verified 2026-05-11
+    // against fal.ai's published Kling-v3 schema (1:1 mirror of Kling's official
+    // API; Kling's own portal at app.klingai.com is Cloudflare-blocked from
+    // server-side fetches). Re-confirm pricing + modelId casing against the
+    // logged-in Kling dashboard before exposing in UI.
+    "kling-v3.0": {
+        modelId: "kling-v3",
+        type: ["text-to-video", "image-to-video"],
+        tier: "T3",
+        fields: {
+            ...klingCommonFields,
+            ...klingI2vFields,
+            // v3.0 widens duration to 3-15s (default 5) per fal.ai schema; override
+            // the v2-era [5, 10] inherited from klingCommonFields. Re-confirm against
+            // direct Kling API docs before fully promising the 3/4/6/7/9/11-15
+            // values in the UI picker (P5.6.b).
+            duration: { allowedValues: [3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15] },
+            mode: { allowedValues: ["std", "pro", "4k"] },
+            // v3.0-only — boolean. NOT the same as v2.6's `audio` pricing-tier toggle.
+            generate_audio: {},
+            // v3.0-only — array of per-shot { prompt, duration?, startFrame? } objects.
+            // Adapter translates canonical `multiShot[]` → Kling `multi_prompt[]` + `shot_type`.
+            multiShot: {},
+        },
+        concurrentRequests: 10,
+        // Verified 2026-05-20 from kling.ai/dev/pricing (V3 tab): std=$0.084/s, pro=$0.112/s, 4k=$0.42/s (no-audio).
+        // With generate_audio=true: std=$0.126/s, pro=$0.168/s (1.5× upcharge); no separate table key yet.
+        cost: {
+            table: {
+                std: { 5: 0.42, 10: 0.84 },
+                pro: { 5: 0.56, 10: 1.12 },
+                "4k": { 5: 2.10, 10: 4.20 },
+            },
+        },
+    },
+    "kling-v2.6": {
+        modelId: "kling-v2-6",
+        type: ["text-to-video", "image-to-video"],
+        tier: "T3",
+        fields: {
+            ...klingCommonFields,
+            ...klingI2vFields,
+            mode: { allowedValues: ["std", "pro"] },
+            // v2.6 `audio` is a pricing-tier toggle, NOT Kling v3's `generate_audio`
+            // request param. Do NOT surface as a "native audio" toggle in UI on v2.x.
             audio: {},
         },
-        concurrentRequests: 3,
+        concurrentRequests: 10,
+        // Verified 2026-05-20 from kling.ai/dev/pricing (V2.6 tab): std no-audio=$0.21/5s, pro no-audio=$0.35/5s.
+        // Audio (pro only): no-voice-control=$0.70/5s; voice-control=$0.84/5s (latter not split here).
+        // Capability map: std mode is no-audio only — `audio.std` intentionally omitted.
         cost: {
-            audio: {
+            noAudio: {
                 table: {
                     std: { 5: 0.21, 10: 0.42 },
-                    pro: { 5: 0.35, 10: 0.7 },
+                    pro: { 5: 0.35, 10: 0.70 },
                 },
             },
-            noAudio: {
+            audio: {
                 table: {
-                    std: { 5: 0.7, 10: 1.4 },
-                    pro: { 5: 0.84, 10: 1.68 },
+                    pro: { 5: 0.70, 10: 1.40 },
                 },
             },
         },
@@ -33,32 +276,31 @@ exports.klingConfigs = {
     "kling-v2.5-turbo": {
         modelId: "kling-v2-5-turbo",
         type: ["text-to-video", "image-to-video"],
+        tier: "T2",
         fields: {
-            prompt: { required: true },
-            aspectRatio: { allowedValues: ["16:9", "9:16", "1:1"] },
-            duration: { allowedValues: [5, 10] },
+            ...klingCommonFields,
+            ...klingI2vFields,
             mode: { allowedValues: ["std", "pro"] },
-            negative_prompt: {},
         },
-        concurrentRequests: 3,
+        concurrentRequests: 10,
+        // Verified 2026-05-20 from kling.ai/dev/pricing (V2.5-Turbo tab): std=$0.21/5s, pro=$0.35/5s.
         cost: {
             table: {
                 std: { 5: 0.21, 10: 0.42 },
-                pro: { 5: 0.35, 10: 0.7 },
+                pro: { 5: 0.35, 10: 0.70 },
             },
         },
     },
     "kling-v2-master": {
         modelId: "kling-v2-master",
         type: ["text-to-video", "image-to-video"],
+        tier: "T3",
         fields: {
-            prompt: { required: true },
-            aspectRatio: { allowedValues: ["16:9", "9:16", "1:1"] },
-            duration: { allowedValues: [5, 10] },
+            ...klingCommonFields,
+            ...klingI2vFields,
             mode: { allowedValues: ["pro"] },
-            negative_prompt: {},
         },
-        concurrentRequests: 3,
+        concurrentRequests: 10,
         cost: {
             table: {
                 pro: { 5: 1.4, 10: 2.8 },
@@ -68,14 +310,14 @@ exports.klingConfigs = {
     "kling-v2.1": {
         modelId: "kling-v2-1",
         type: ["image-to-video"],
+        tier: "T2",
         fields: {
-            prompt: { required: true },
-            aspectRatio: { allowedValues: ["16:9", "9:16", "1:1"] },
-            duration: { allowedValues: [5, 10] },
+            ...klingCommonFields,
+            ...klingI2vFields,
+            inputImageUrl: { required: true },
             mode: { allowedValues: ["std", "pro"] },
-            negative_prompt: {},
         },
-        concurrentRequests: 3,
+        concurrentRequests: 10,
         cost: {
             table: {
                 std: { 5: 0.28, 10: 0.56 },
@@ -86,14 +328,13 @@ exports.klingConfigs = {
     "kling-v2.1-master": {
         modelId: "kling-v2-1-master",
         type: ["text-to-video", "image-to-video"],
+        tier: "T3",
         fields: {
-            prompt: { required: true },
-            aspectRatio: { allowedValues: ["16:9", "9:16", "1:1"] },
-            duration: { allowedValues: [5, 10] },
+            ...klingCommonFields,
+            ...klingI2vFields,
             mode: { allowedValues: ["pro"] },
-            negative_prompt: {},
         },
-        concurrentRequests: 3,
+        concurrentRequests: 10,
         cost: {
             table: {
                 pro: { 5: 1.4, 10: 2.8 },
@@ -104,23 +345,25 @@ exports.klingConfigs = {
     // Endpoint: POST /v1/videos/avatar/image2video
     // Requires: image (avatar face URL) + sound_file (audio URL)
     // Duration is determined by audio file length — NOT a fixed selection
-    // TODO: confirm exact per-second pricing from Kling dashboard
     "kling-avatar": {
         modelId: "kling-avatar",
         type: ["avatar-video"],
+        tier: "T2",
         fields: {
             inputImageUrl: { required: true },
             inputAudioUrl: {},
             mode: { allowedValues: ["std", "pro"], default: "std" },
             prompt: {},
+            callbackUrl: {},
+            externalTaskId: {},
         },
-        concurrentRequests: 3,
+        concurrentRequests: 5,
         cost: {
-            // per-second pricing by mode — read via cost.perMode[mode].perSecond
-            // TODO: confirm exact per-second pricing from Kling dashboard
+            // Verified 2026-05-20 from kling.ai/dev/pricing (Functional Models tab):
+            // std=$0.056/s (0.4 units/s), pro=$0.112/s (0.8 units/s).
             perMode: {
-                std: { perSecond: 0.014 },
-                pro: { perSecond: 0.028 },
+                std: { perSecond: 0.056 },
+                pro: { perSecond: 0.112 },
             },
         },
         avatar: {

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

@@ -1 +1 @@
-{"version":3,"file":"minimax.d.ts","sourceRoot":"","sources":["../../../../src/globals/aiModels/providers/minimax.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,WAAW,EAAE,MAAM,UAAU,CAAC;AACvC,OAAO,EAAE,YAAY,EAAE,MAAM,UAAU,CAAC;AAExC,eAAO,MAAM,cAAc,EAAE,OAAO,CAAC,MAAM,CAAC,WAAW,EAAE,YAAY,CAAC,CAgErE,CAAC"}
+{"version":3,"file":"minimax.d.ts","sourceRoot":"","sources":["../../../../src/globals/aiModels/providers/minimax.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,WAAW,EAAE,MAAM,UAAU,CAAC;AACvC,OAAO,EAAE,YAAY,EAAE,MAAM,UAAU,CAAC;AAExC,eAAO,MAAM,cAAc,EAAE,OAAO,CAAC,MAAM,CAAC,WAAW,EAAE,YAAY,CAAC,CA0IrE,CAAC"}