// Protocol capabilities, not prices. Keep defaults and validation shared by // native requests, compatibility protocols, and submission billing facts. const WAN_MODELS = { "wan3.0-video": { kind: "all", resolutions: ["480P", "720P", "1080P"], defaultResolution: "1080P", maxDuration: 30 }, "wan3.0-video-prime": { kind: "all", resolutions: ["480P", "720P", "1080P"], defaultResolution: "1080P", maxDuration: 30 }, "wan2.7-t2v": { kind: "t2v", resolutions: ["720P", "1080P"], defaultResolution: "1080P", maxDuration: 15 }, "wan2.7-i2v": { kind: "media", resolutions: ["720P", "1080P"], defaultResolution: "1080P", maxDuration: 15 }, "wan2.6-t2v": { kind: "size", resolutions: ["720P", "1080P"], defaultResolution: "1080P", maxDuration: 15 }, "wan2.6-t2v-us": { kind: "size", resolutions: ["720P", "1080P"], defaultResolution: "1080P", durations: [5, 10, 15] }, "wan2.6-i2v": { kind: "image", resolutions: ["720P", "1080P"], defaultResolution: "1080P", maxDuration: 15 }, "wan2.6-i2v-flash": { kind: "image", resolutions: ["720P", "1080P"], defaultResolution: "1080P", maxDuration: 15 }, "wan2.6-i2v-us": { kind: "image", resolutions: ["720P", "1080P"], defaultResolution: "1080P", durations: [5, 10, 15] }, "wan2.5-t2v-preview": { kind: "size", resolutions: ["480P", "720P", "1080P"], defaultResolution: "1080P", durations: [5, 10] }, "wan2.5-i2v-preview": { kind: "image", resolutions: ["480P", "720P", "1080P"], defaultResolution: "1080P", durations: [5, 10] }, "wan2.2-t2v-plus": { kind: "size", resolutions: ["480P", "1080P"], defaultResolution: "1080P", durations: [5] }, "wan2.2-i2v-flash": { kind: "image", resolutions: ["480P", "720P", "1080P"], defaultResolution: "720P", durations: [5] }, "wan2.2-i2v-plus": { kind: "image", resolutions: ["480P", "1080P"], defaultResolution: "1080P", durations: [5] }, "wan2.2-kf2v-flash": { kind: "frames", resolutions: ["480P", "720P", "1080P"], defaultResolution: "720P", durations: [5] }, "wan2.2-s2v": { kind: "speech", resolutions: ["480P", "720P"], defaultResolution: "480P" }, "wanx2.1-t2v-plus": { kind: "size", resolutions: ["720P"], defaultResolution: "720P", durations: [5] }, "wanx2.1-t2v-turbo": { kind: "size", resolutions: ["480P", "720P"], defaultResolution: "720P", durations: [5] }, "wanx2.1-i2v-plus": { kind: "image", resolutions: ["720P"], defaultResolution: "720P", durations: [5] }, "wanx2.1-i2v-turbo": { kind: "image", resolutions: ["480P", "720P"], defaultResolution: "720P", durations: [3, 4, 5] }, }; // HTTP contracts (SDK synchronous calls can also wrap polling): // https://help.aliyun.com/zh/model-studio/text-to-image-v2-api-reference // https://help.aliyun.com/zh/model-studio/wan-image-generation-api-reference // https://help.aliyun.com/zh/model-studio/wan-image-generation-and-editing-api-reference // https://help.aliyun.com/zh/model-studio/wan2-5-image-edit-api-reference // https://help.aliyun.com/zh/model-studio/wanx-image-edit-api-reference // https://help.aliyun.com/zh/model-studio/qwen-image-api // https://help.aliyun.com/zh/model-studio/qwen-image-edit-api // https://help.aliyun.com/zh/model-studio/qwen-image-generation-and-editing-api-reference // https://help.aliyun.com/zh/model-studio/z-image-api-reference // Kinds: image27/image26/t2i use multimodal-generation (sync) or // image-generation/generation (async) with output.choices; legacy uses the // async text2image/image-synthesis with output.results; legacy_edit uses the // async image2image/image-synthesis with input.images and output.results; // legacy_imageedit (wanx2.1-imageedit) uses the same async // image2image/image-synthesis with input.function, input.base_image_url and an // optional input.mask_image_url; qwen (Qwen-Image, Qwen-Image-Edit, Z-Image) // uses multimodal-generation synchronously only, with output.choices. const IMAGE_MODELS = { "wan2.7-image-pro": "image27", "wan2.7-image": "image27", "wan2.6-image": "image26", "wan2.6-t2i": "t2i", "wan2.5-t2i-preview": "legacy", "wan2.2-t2i-flash": "legacy", "wan2.2-t2i-plus": "legacy", "wanx2.1-t2i-turbo": "legacy", "wanx2.1-t2i-plus": "legacy", "wanx2.0-t2i-turbo": "legacy", "wan2.5-i2i-preview": "legacy_edit", "wanx2.1-imageedit": "legacy_imageedit", "qwen-image": "qwen", "qwen-image-plus": "qwen", "qwen-image-max": "qwen", "qwen-image-2.0": "qwen", "qwen-image-2.0-pro": "qwen", "qwen-image-3.0": "qwen", "qwen-image-3.0-pro": "qwen", "qwen-image-edit": "qwen", "qwen-image-edit-plus": "qwen", "qwen-image-edit-max": "qwen", "z-image-turbo": "qwen", }; // wanx2.1-imageedit selects its editing operation with input.function. The // parameters below belong to specific functions and are forwarded as // documented; none of them changes the billed image count. const IMAGE_EDIT_FUNCTIONS = [ "stylization_all", "stylization_local", "description_edit", "description_edit_with_mask", "remove_watermark", "expand", "super_resolution", "colorization", "doodle", "control_cartoon_feature", ]; const IMAGE_EDIT_FUNCTION_PARAMETERS = ["strength", "top_scale", "bottom_scale", "left_scale", "right_scale", "upscale_factor", "is_sketch"]; // Documented limits of the multimodal-generation image models: the output // count range and the number of input images (minInput 1 marks edit-only // models). Fixed-count models reject n > 1 upstream ("n must be 1"). const QWEN_IMAGE_LIMITS = { "qwen-image": { maxN: 1, maxInput: 0 }, "qwen-image-plus": { maxN: 1, maxInput: 0 }, "qwen-image-max": { maxN: 1, maxInput: 0 }, "qwen-image-2.0": { maxN: 6, maxInput: 3 }, "qwen-image-2.0-pro": { maxN: 6, maxInput: 3 }, "qwen-image-3.0": { maxN: 6, maxInput: 3 }, "qwen-image-3.0-pro": { maxN: 6, maxInput: 3 }, "qwen-image-edit": { maxN: 1, maxInput: 3, minInput: 1 }, "qwen-image-edit-plus": { maxN: 6, maxInput: 3, minInput: 1 }, "qwen-image-edit-max": { maxN: 6, maxInput: 3, minInput: 1 }, "z-image-turbo": { maxN: 1, maxInput: 0 }, }; // Dated snapshots resolve to the undated profile through modelKey(). const IMAGE_MODEL_SNAPSHOTS = [ "qwen-image-plus-2026-01-09", "qwen-image-max-2025-12-30", "qwen-image-2.0-2026-03-03", "qwen-image-2.0-pro-2026-03-03", "qwen-image-2.0-pro-2026-04-22", "qwen-image-2.0-pro-2026-06-22", "qwen-image-edit-plus-2025-10-30", "qwen-image-edit-plus-2025-12-15", "qwen-image-edit-max-2026-01-16", ]; const QWEN_IMAGE3_MODELS = ["qwen-image-3.0", "qwen-image-3.0-pro"]; const Z_IMAGE_MODELS = ["z-image-turbo"]; // Qwen-Image-3.0 usage: output_image_type / input_image_type switch to the // 2K tier above this pixel area. const QWEN_IMAGE3_TIER_MAX_PIXELS = 2250000; // Largest documented per-request image output (wan2.7 group generation). const MAX_IMAGE_OUTPUTS = 12; // Legacy per-call pricing multiplied Z-Image requests with prompt rewriting // by this ratio; task expressions read the prompt_extend fact instead. const Z_IMAGE_PROMPT_EXTEND_RATIO = 2; // Base64 uploads on the OpenAI edits endpoint are bounded by DashScope's 10 MB input image limit. const MAX_INPUT_IMAGE_BYTES = 10485760; const IMAGE_UNIT_LABEL = { en: "image", zh: "张", "zh-TW": "張", fr: "image", ja: "枚", ru: "изображение", vi: "ảnh" }; const IMAGE_COUNT_FIELD = { type: "number", unit: "count", unitLabel: IMAGE_UNIT_LABEL, description: { en: "Image generation unit price", zh: "图片生成单价" }, }; const WAN_IMAGE_USAGE_SCHEMA = { image_count: IMAGE_COUNT_FIELD }; // Z-Image is priced differently when prompt rewriting is enabled. const Z_IMAGE_USAGE_SCHEMA = { image_count: IMAGE_COUNT_FIELD, prompt_extend: { type: "boolean", description: { en: "Whether prompt rewriting is enabled", zh: "是否开启提示词改写" }, }, }; // Qwen-Image-3.0 meters output images by tier and input images separately. // Estimated at submit from the requested size and input count; settled from // usage.output_image_count, usage.output_image_type and usage.input_image_count. const QWEN_IMAGE3_USAGE_SCHEMA = { image_count: IMAGE_COUNT_FIELD, output_image_type: { enum: ["qima_output_1k", "qima_output_2k"], enumLabels: { qima_output_1k: { en: "1K output", zh: "1K 输出" }, qima_output_2k: { en: "2K output", zh: "2K 输出" } }, description: { en: "Output image tier", zh: "输出图片档位" }, }, input_image_count: { type: "number", unit: "count", unitLabel: IMAGE_UNIT_LABEL, description: { en: "Input image unit price", zh: "输入图片单价" }, }, }; // Wan video usage facts. The resolution options follow the model's // capability table so pricing lists only the tiers the model offers. function wanVideoUsageSchema(resolutions, audio) { const resolutionLabels = {}; for (const resolution of resolutions) resolutionLabels[resolution] = { en: resolution, zh: resolution }; const schema = { // Billable seconds: output duration, or input + output for Wan3. seconds: { type: "number", unit: "second", description: { en: "Video generation unit price", zh: "视频生成单价" }, }, // Requested output video resolution. resolution: { enum: resolutions, enumLabels: resolutionLabels, description: { en: "Output video resolution", zh: "输出视频分辨率" }, }, }; // wan2.6-i2v-flash prices audio and silent output differently. parameters.audio // defaults to true; the completion usage.audio flag replaces the estimate. if (audio) { schema.audio = { type: "boolean", description: { en: "Whether audio is generated", zh: "是否生成音频" }, }; } return schema; } const WAN_AUDIO_VIDEO_MODELS = ["wan2.6-i2v-flash"]; // Dated snapshots share the undated model's capabilities through modelKey. const WAN_VIDEO_SNAPSHOTS = ["wan2.7-t2v-2026-04-25", "wan2.7-t2v-2026-06-12", "wan2.7-i2v-2026-04-25"]; const WAN_VIDEO_RESOLUTIONS = ["480P", "720P", "1080P"]; // One usage profile per distinct resolution set and audio option. function wanVideoUsageProfiles() { const profiles = []; for (const model of Object.keys(WAN_MODELS).concat(WAN_VIDEO_SNAPSHOTS)) { const key = modelKey(model); const resolutions = WAN_MODELS[key].resolutions; const audio = WAN_AUDIO_VIDEO_MODELS.includes(key); const shape = resolutions.join("/") + (audio ? "+audio" : ""); let profile = profiles.find((entry) => entry.shape === shape); if (!profile) { profile = { shape: shape, models: [], schema: wanVideoUsageSchema(resolutions, audio) }; profiles.push(profile); } profile.models.push(model); } return profiles.map((profile) => ({ models: profile.models, schema: profile.schema })); } export const meta = { apiVersion: 1, key: "alibaba", name: "Alibaba Bailian", icon: "Bailian.Color", description: { en: "Alibaba Cloud Bailian image and video generation (Wan, Qwen-Image, Z-Image)", zh: "阿里云百炼图片与视频生成（万相、千问图像、Z-Image）", }, version: "1.4.2", author: { name: "QuantumNous" }, channelTypes: [17], // Literal metadata also supports the dashboard's static script preview. models: [ "wan3.0-video", "wan3.0-video-prime", "wan2.7-t2v", "wan2.7-t2v-2026-04-25", "wan2.7-t2v-2026-06-12", "wan2.7-i2v", "wan2.7-i2v-2026-04-25", "wan2.6-t2v", "wan2.6-t2v-us", "wan2.6-i2v", "wan2.6-i2v-flash", "wan2.6-i2v-us", "wan2.5-t2v-preview", "wan2.5-i2v-preview", "wan2.2-t2v-plus", "wan2.2-i2v-flash", "wan2.2-i2v-plus", "wan2.2-kf2v-flash", "wan2.2-s2v", "wanx2.1-t2v-plus", "wanx2.1-t2v-turbo", "wanx2.1-i2v-plus", "wanx2.1-i2v-turbo", "wan2.7-image-pro", "wan2.7-image", "wan2.6-image", "wan2.6-t2i", "wan2.5-t2i-preview", "wan2.2-t2i-flash", "wan2.2-t2i-plus", "wanx2.1-t2i-turbo", "wanx2.1-t2i-plus", "wanx2.0-t2i-turbo", "wan2.5-i2i-preview", "wanx2.1-imageedit", "qwen-image", "qwen-image-plus", "qwen-image-max", "qwen-image-2.0", "qwen-image-2.0-pro", "qwen-image-3.0", "qwen-image-3.0-pro", "qwen-image-edit", "qwen-image-edit-plus", "qwen-image-edit-max", "z-image-turbo", ].concat(IMAGE_MODEL_SNAPSHOTS), fetchMode: "per_task", upstreams: ["vendor", "new_api"], submitResponseTypes: ["json", "sse"], requiredCapabilities: ["duration-auto@1"], usageSchema: { ...WAN_IMAGE_USAGE_SCHEMA, ...wanVideoUsageSchema(WAN_VIDEO_RESOLUTIONS, false) }, usageProfiles: [ { models: Object.keys(IMAGE_MODELS) .concat(IMAGE_MODEL_SNAPSHOTS) .filter((name) => !QWEN_IMAGE3_MODELS.includes(name) && !Z_IMAGE_MODELS.includes(name)), schema: WAN_IMAGE_USAGE_SCHEMA, }, { models: Z_IMAGE_MODELS, schema: Z_IMAGE_USAGE_SCHEMA }, { models: QWEN_IMAGE3_MODELS, schema: QWEN_IMAGE3_USAGE_SCHEMA }, ].concat(wanVideoUsageProfiles()), routes: [ // Synchronous vendor call with no task to re-query; the response is // delivered once and never persisted. { method: "POST", path: "/ali/api/v1/services/aigc/multimodal-generation/generation", type: "submit", models: ["wan2.7-image-pro", "wan2.7-image", "wan2.6-image", "wan2.6-t2i"].concat(Object.keys(QWEN_IMAGE_LIMITS), IMAGE_MODEL_SNAPSHOTS), decode: "createImageTask", render: "imageCreated", retainResult: false, }, { method: "POST", path: "/ali/api/v1/services/aigc/image-generation/generation", type: "submit", models: ["wan2.7-image-pro", "wan2.7-image", "wan2.6-image", "wan2.6-t2i"], decode: "createImageTask", render: "taskCreated", }, { method: "POST", path: "/ali/api/v1/services/aigc/text2image/image-synthesis", type: "submit", models: ["wan2.5-t2i-preview", "wan2.2-t2i-flash", "wan2.2-t2i-plus", "wanx2.1-t2i-turbo", "wanx2.1-t2i-plus", "wanx2.0-t2i-turbo"], decode: "createImageTask", render: "taskCreated", }, { method: "POST", path: "/ali/api/v1/services/aigc/image2image/image-synthesis", type: "submit", models: ["wan2.5-i2i-preview", "wanx2.1-imageedit"], decode: "createImageTask", render: "taskCreated", }, { method: "POST", path: "/ali/api/v1/services/aigc/video-generation/video-synthesis", type: "submit", decode: "createVideoTask", render: "taskCreated" }, { method: "POST", path: "/ali/api/v1/services/aigc/image2video/video-synthesis", type: "submit", decode: "createVideoTask", render: "taskCreated" }, { method: "GET", path: "/ali/api/v1/tasks/:task_id", type: "query", render: "taskStatus" }, ], protocols: [ { name: "openai_responses", supports: ["stream", "sync", "background"] }, { name: "openai_video", models: Object.keys(WAN_MODELS).concat(["wan2.7-t2v-2026-04-25", "wan2.7-t2v-2026-06-12", "wan2.7-i2v-2026-04-25"]) }, { name: "openai_image", models: Object.keys(IMAGE_MODELS).concat(IMAGE_MODEL_SNAPSHOTS) }, ], }; function trimmed(value) { return String(value || "").trim(); } function viaNewAPI(ctx) { return !!(ctx.upstream && ctx.upstream.kind === "new_api"); } // Another New API gateway serves the DashScope wire format only on this // plugin's /ali native routes; DashScope itself serves the unprefixed paths. function apiRoot(ctx) { return ctx.baseUrl + (viaNewAPI(ctx) ? "/ali" : ""); } function firstImage(req) { if (trimmed(req.image)) return trimmed(req.image); for (const image of req.images || []) if (trimmed(image)) return trimmed(image); return trimmed(req.input_reference); } function secondImage(req) { let count = 0; for (const image of req.images || []) { if (!trimmed(image)) continue; count++; if (count === 2) return trimmed(image); } return ""; } const LEGACY_SIZES = { "480P": { "16:9": "832*480", "9:16": "480*832", "1:1": "624*624" }, "720P": { "16:9": "1280*720", "9:16": "720*1280", "1:1": "960*960", "4:3": "1088*832", "3:4": "832*1088" }, "1080P": { "16:9": "1920*1080", "9:16": "1080*1920", "1:1": "1440*1440", "4:3": "1632*1248", "3:4": "1248*1632" }, }; // Wan2.7 changed the pixel sizes for 4:3 and 3:4. Legacy sizes remain valid // compatibility inputs, but never replace the target model's native sizes. const MODERN_SIZES = { "720P": { "16:9": "1280*720", "9:16": "720*1280", "1:1": "960*960", "4:3": "1104*832", "3:4": "832*1104" }, "1080P": { "16:9": "1920*1080", "9:16": "1080*1920", "1:1": "1440*1440", "4:3": "1648*1248", "3:4": "1248*1648" }, }; function modelKey(model) { return String(model || "") .replace(/-\d{4}-\d{2}-\d{2}$/, "") .replace(/^wan2\.1-/, "wanx2.1-"); } function modelProfile(model) { const key = modelKey(model); if (!Object.prototype.hasOwnProperty.call(WAN_MODELS, key)) throw new Error("unsupported Wan model: " + model); return WAN_MODELS[key]; } function imageModel(ctx) { const key = modelKey(ctx.upstreamModel || ctx.model || (ctx.requestBody || {}).model); return Object.prototype.hasOwnProperty.call(IMAGE_MODELS, key) ? IMAGE_MODELS[key] : undefined; } function qwenImage3(ctx) { return QWEN_IMAGE3_MODELS.includes(modelKey(ctx.upstreamModel || ctx.model || (ctx.requestBody || {}).model)); } function zImage(ctx) { return Z_IMAGE_MODELS.includes(modelKey(ctx.upstreamModel || ctx.model || (ctx.requestBody || {}).model)); } // Host file placeholders stand in for uploaded edit images; the host inlines // them as data URLs before the request is sent. function isFileRef(value) { return !!value && typeof value === "object" && !Array.isArray(value) && typeof value.__fileRef === "string"; } function isImageInput(value) { return isFileRef(value) || (typeof value === "string" && /^(https?:\/\/|data:image\/)/i.test(value)); } function sizePixels(size) { const match = /^(\d+)\*(\d+)$/.exec(size); if (!match) return null; const width = Number(match[1]), height = Number(match[2]); if (!Number.isSafeInteger(width) || !Number.isSafeInteger(height) || width <= 0 || height <= 0) return null; return { width: width, height: height, pixels: width * height }; } function convertImage(ctx) { const req = ctx.requestBody || {}; const model = ctx.upstreamModel || req.model; const profile = imageModel(ctx); if (!profile) throw new Error("unsupported image model: " + model); const limits = QWEN_IMAGE_LIMITS[modelKey(model)]; const metadata = objectValue(req.metadata, "metadata"); if (metadata.model !== undefined && metadata.model !== model) throw new Error("can't change model with metadata"); const mode = metadata.upstream_mode ?? (profile === "qwen" ? "sync" : "async"); if (mode !== "sync" && mode !== "async") throw new Error("upstream_mode must be sync or async"); const legacyProfile = profile === "legacy" || profile === "legacy_edit" || profile === "legacy_imageedit"; if (legacyProfile && mode === "sync") throw new Error("this image model only supports asynchronous HTTP calls"); if (profile === "qwen" && mode === "async") throw new Error("this image model only supports synchronous HTTP calls"); const parameters = {}; for (const key of [ "n", "size", "negative_prompt", "prompt_extend", "prompt_extend_mode", "enable_thinking", "watermark", "seed", "enable_interleave", "max_images", "enable_sequential", "thinking_mode", "bbox_list", "color_palette", ]) { if (req[key] !== undefined) parameters[key] = req[key]; } if (profile === "legacy_imageedit") { for (const key of IMAGE_EDIT_FUNCTION_PARAMETERS) if (req[key] !== undefined) parameters[key] = req[key]; } Object.assign(parameters, objectValue(metadata.parameters, "metadata.parameters")); for (const key of ["prompt_extend", "enable_thinking", "watermark", "enable_interleave", "enable_sequential", "thinking_mode", "stream"]) { if (parameters[key] !== undefined && typeof parameters[key] !== "boolean") throw new Error(key + " must be a boolean"); } if (parameters.stream === true && (mode !== "sync" || profile !== "image26" || !parameters.enable_interleave)) throw new Error("upstream streaming requires synchronous wan2.6-image interleaved output"); if (parameters.enable_interleave && profile !== "image26") throw new Error("enable_interleave is only supported by wan2.6-image"); if (parameters.enable_sequential && profile !== "image27") throw new Error("enable_sequential is only supported by wan2.7-image models"); if (parameters.enable_interleave && mode === "sync") parameters.stream = true; const maxN = parameters.enable_interleave ? 1 : parameters.enable_sequential ? 12 : profile === "qwen" ? limits.maxN : 4; parameters.n = parameters.n ?? (mode === "sync" || parameters.enable_interleave || profile === "qwen" || profile === "legacy_imageedit" ? 1 : profile === "image27" ? parameters.enable_sequential ? 12 : 1 : 4); // Provider limits are stricter than the host's dto.MaxImageN count ceiling. // Fixed-count models are rejected here instead of letting DashScope answer // "n must be 1" after the quantity was reserved. if (!Number.isInteger(parameters.n) || parameters.n < 1 || parameters.n > maxN) throw new Error(maxN === 1 ? "n must be 1 for this model" : "n must be an integer between 1 and " + maxN); if (parameters.max_images !== undefined && (!Number.isInteger(parameters.max_images) || parameters.max_images < 1 || parameters.max_images > 5)) throw new Error("max_images must be an integer between 1 and 5"); if (parameters.enable_interleave) parameters.max_images = parameters.max_images ?? 5; if (parameters.seed !== undefined && (!Number.isInteger(parameters.seed) || parameters.seed < 0 || parameters.seed > 2147483647)) throw new Error("seed must be an integer between 0 and 2147483647"); if (parameters.prompt_extend_mode !== undefined && parameters.prompt_extend_mode !== "direct" && parameters.prompt_extend_mode !== "agent") throw new Error("prompt_extend_mode must be direct or agent"); const input = Object.assign({}, objectValue(metadata.input, "metadata.input")); let listedImages = req.images; if (listedImages === undefined) { const first = isFileRef(req.image) ? req.image : firstImage(req); listedImages = first ? [first] : []; } if (!Array.isArray(listedImages)) throw new Error("images must be an array"); let imageCount = 0; if (legacyProfile) { input.prompt = input.prompt ?? req.prompt; if (typeof input.prompt !== "string" || !input.prompt.trim()) throw new Error("prompt is required"); if (input.messages !== undefined) throw new Error("this model does not accept input.messages"); if (parameters.negative_prompt !== undefined) { if (profile === "legacy_imageedit") throw new Error("this model does not accept negative_prompt"); input.negative_prompt = parameters.negative_prompt; delete parameters.negative_prompt; } if (profile === "legacy") { if (listedImages.length) throw new Error("this model only supports text-to-image input"); } else if (profile === "legacy_edit") { input.images = input.images ?? listedImages; if (!Array.isArray(input.images) || input.images.length < 1 || input.images.length > 3 || !input.images.every(isImageInput)) throw new Error("this model requires 1 to 3 input images as HTTP URLs or Base64 data URLs"); imageCount = input.images.length; } else { // wanx2.1-imageedit takes one base image and selects the operation with // input.function; an OpenAI edit defaults to instruction editing, or to // local repainting when a mask is supplied. if (input.images !== undefined) throw new Error("this model takes input.base_image_url, not input.images"); if (input.base_image_url === undefined && listedImages.length === 1) input.base_image_url = listedImages[0]; if (listedImages.length > 1 || !isImageInput(input.base_image_url)) throw new Error("this model requires exactly one input image as an HTTP URL or Base64 data URL"); if (input.mask_image_url === undefined && req.mask !== undefined) input.mask_image_url = req.mask; input.function = input.function ?? req.function ?? (input.mask_image_url !== undefined ? "description_edit_with_mask" : "description_edit"); if (!IMAGE_EDIT_FUNCTIONS.includes(input.function)) throw new Error("function must be one of " + IMAGE_EDIT_FUNCTIONS.join(", ")); if (input.function === "description_edit_with_mask" && !isImageInput(input.mask_image_url)) throw new Error("description_edit_with_mask requires a mask image as an HTTP URL or Base64 data URL"); if (input.function !== "description_edit_with_mask" && input.mask_image_url !== undefined) throw new Error("mask images are only used by description_edit_with_mask"); if (parameters.size !== undefined) throw new Error("this model does not accept size"); imageCount = 1; } } else if (input.messages === undefined) { const content = []; for (const image of listedImages) content.push({ image: image }); content.push({ text: req.prompt }); input.messages = [{ role: "user", content: content }]; } if (!legacyProfile) { if ( !Array.isArray(input.messages) || input.messages.length !== 1 || !input.messages[0] || input.messages[0].role !== "user" || !Array.isArray(input.messages[0].content) ) throw new Error("input.messages must contain one user message with a content array"); let texts = 0; for (const part of input.messages[0].content) { if (!part || typeof part !== "object" || Array.isArray(part)) throw new Error("invalid image message content"); if (part.text !== undefined) { if (typeof part.text !== "string" || !part.text.trim()) throw new Error("text must be a non-empty string"); texts++; } if (part.image !== undefined) { if (!isImageInput(part.image)) throw new Error("image must be an HTTP URL or Base64 data URL"); imageCount++; } } if (texts !== 1) throw new Error("input.messages must contain exactly one text prompt"); const maxInput = profile === "image27" ? 9 : profile === "image26" ? (parameters.enable_interleave ? 1 : 4) : profile === "qwen" ? limits.maxInput : 0; if (imageCount > maxInput) throw new Error(maxInput === 0 ? "this model only supports text-to-image input" : "too many input images for this model"); if (profile === "image26" && !parameters.enable_interleave && imageCount === 0) throw new Error("wan2.6-image editing requires a reference image; use wan2.6-t2i for text-to-image"); if (profile === "qwen" && limits.minInput && imageCount < limits.minInput) throw new Error("this image editing model requires at least one input image"); } if (parameters.size !== undefined) { if (typeof parameters.size !== "string") throw new Error("size must be a string"); parameters.size = parameters.size.replace(/x/i, "*"); const supports4K = modelKey(model) === "wan2.7-image-pro" && !imageCount && !parameters.enable_sequential; if (["1K", "2K", "4K"].includes(parameters.size)) { if ((profile !== "image26" && profile !== "image27") || parameters.enable_interleave) throw new Error("this model requires a width*height image size"); if (parameters.size === "4K" && !supports4K) throw new Error("4K is only supported by wan2.7-image-pro text-to-image without sequential output"); } else { const dims = sizePixels(parameters.size); if (!dims) throw new Error("size must be width*height or a supported resolution preset"); const width = dims.width, height = dims.height; const legacySize = profile === "legacy" && modelKey(model) !== "wan2.5-t2i-preview"; const maxPixels = profile === "image27" ? supports4K ? 4096 * 4096 : 2048 * 2048 : profile === "image26" ? parameters.enable_interleave ? 1280 * 1280 : 2048 * 2048 : profile === "qwen" ? 2048 * 2048 : profile === "legacy_edit" ? 1280 * 1280 : 1440 * 1440; const minPixels = profile === "qwen" ? 512 * 512 : 0; const ratio = profile === "image27" || profile === "qwen" ? 8 : 4; if ( legacySize ? width < 512 || height < 512 || width > 1440 || height > 1440 : dims.pixels > maxPixels || dims.pixels < minPixels || width / height > ratio || height / width > ratio ) throw new Error("size is outside the model's pixel and aspect-ratio limits"); } } const service = profile === "legacy" ? "text2image/image-synthesis" : profile === "legacy_edit" || profile === "legacy_imageedit" ? "image2image/image-synthesis" : mode === "sync" ? "multimodal-generation/generation" : "image-generation/generation"; return { body: { model: model, input: input, parameters: parameters }, service: service, synchronous: mode === "sync", action: imageCount ? "image_to_image" : "text_to_image", inputImages: imageCount, }; } // Submit-time billing facts: the requested count (max_images for interleaved // output), the Z-Image prompt rewriting flag, and the Qwen-Image-3.0 tier // estimated from the requested size (an unspecified size reserves the higher // tier). Completion facts replace these key by key. function imageEstimate(ctx, converted) { const parameters = converted.body.parameters; const facts = { image_count: parameters.enable_interleave ? parameters.max_images : parameters.n }; if (zImage(ctx)) facts.prompt_extend = parameters.prompt_extend === true; if (qwenImage3(ctx)) { const dims = typeof parameters.size === "string" ? sizePixels(parameters.size) : null; facts.output_image_type = dims && dims.pixels <= QWEN_IMAGE3_TIER_MAX_PIXELS ? "qima_output_1k" : "qima_output_2k"; facts.input_image_count = converted.inputImages; } return facts; } // Legacy per-call pricing multiplies the model price by these ratios; the // Z-Image prompt rewriting surcharge survives as its own ratio because the // boolean fact cannot be a multiplier. function imageRatios(ctx, converted, count) { const ratios = { image_count: count }; if (zImage(ctx) && converted.body.parameters.prompt_extend === true) ratios.prompt_extend_ratio = Z_IMAGE_PROMPT_EXTEND_RATIO; return ratios; } function imageContent(body) { const output = (body && body.output) || {}; const content = []; for (const result of Array.isArray(output.results) ? output.results : []) { if (result && typeof result.url === "string" && result.url) content.push({ image: result.url }); } for (const choice of Array.isArray(output.choices) ? output.choices : []) { for (const part of choice && choice.message && Array.isArray(choice.message.content) ? choice.message.content : []) { if (part && typeof part.image === "string" && part.image) content.push({ image: part.image }); else if (part && typeof part.text === "string") content.push({ text: part.text }); } } return content; } // Completion facts. The count prefers usage.image_count, then the // Qwen-Image-3.0 usage.output_image_count, and otherwise counts image payloads // flattened across output.results[] and every choices[].message.content[] // part; payload-less entries are never counted. An inconsistent upstream // count is rejected so the host keeps the reservation instead of guessing. function imageUsage(ctx, body) { const usage = (body && body.usage) || {}; const content = imageContent(body); const images = content.filter(function (part) { return part.image; }); if (images.length > MAX_IMAGE_OUTPUTS) throw new Error("too many output images"); const reported = usage.image_count !== undefined ? usage.image_count : usage.output_image_count; let count; if (reported !== undefined) { // Do not let fractional, negative or oversized upstream counts change billing. if (!Number.isInteger(reported) || reported < 0 || reported > MAX_IMAGE_OUTPUTS || (images.length && reported === 0)) throw new Error("invalid upstream image count"); count = reported; } else if (content.length) count = images.length; else return {}; const facts = { image_count: count }; if (qwenImage3(ctx)) { if (QWEN_IMAGE3_USAGE_SCHEMA.output_image_type.enum.includes(usage.output_image_type)) facts.output_image_type = usage.output_image_type; if (Number.isInteger(usage.input_image_count) && usage.input_image_count >= 0 && usage.input_image_count <= 3) facts.input_image_count = usage.input_image_count; } return facts; } // One OpenAI ImageResponse entry from a DashScope image payload. function imageDatum(image) { if (/^https?:\/\//i.test(image)) return { url: image }; const match = /^data:[^;,]*;base64,(.*)$/s.exec(image); return { b64_json: match ? match[1] : image }; } function objectValue(value, name) { if (value === undefined || value === null) return {}; if (typeof value !== "object" || Array.isArray(value)) throw new Error(name + " must be an object"); return value; } // Decoders fold wan3.0's "-1 = smart duration" sentinel, wherever the client // put it, into one boolean marker; convert sends -1 upstream and extractUsage // reserves 30 seconds, the longest wan3.0 output, for it. function normalizeRequest(value) { const req = Object.assign({}, value); const metadata = objectValue(req.metadata, "metadata"); const parameters = Object.assign({}, objectValue(metadata.parameters, "metadata.parameters")); if (req.metadata !== undefined) { req.metadata = Object.assign({}, metadata, { parameters: parameters }); objectValue(metadata.input, "metadata.input"); } const duration = parameters.duration ?? req.duration ?? req.seconds ?? (req.auto_duration === true ? -1 : undefined); for (const key of ["duration", "seconds"]) { if (Number(req[key]) === -1) delete req[key]; } if (Number(parameters.duration) === -1) delete parameters.duration; if (Number(duration) === -1) { req.auto_duration = true; delete req.duration; delete req.seconds; } else delete req.auto_duration; // Normalize enum facts for host validation against the model's usage schema. if (req.resolution != null) req.resolution = normalizeResolution(req.resolution); if (parameters.resolution != null) parameters.resolution = normalizeResolution(parameters.resolution); return req; } function normalizeResolution(value) { let resolution = trimmed(value).toUpperCase(); if (!resolution) return ""; if (!resolution.endsWith("P")) resolution += "P"; return resolution; } function videoSize(value) { const size = trimmed(value).replace(/x/i, "*"); if (!size.includes("*")) return { resolution: normalizeResolution(size) }; for (const sizes of [LEGACY_SIZES, MODERN_SIZES]) { for (const resolution of Object.keys(sizes)) { for (const ratio of Object.keys(sizes[resolution])) { if (sizes[resolution][ratio] === size) return { resolution: resolution, ratio: ratio }; } } } throw new Error("invalid size: " + size); } function videoAction(req) { const input = objectValue((req.metadata || {}).input, "metadata.input"); for (const source of [req, input]) { if ( firstImage(source) || trimmed(source.img_url) || trimmed(source.image_url) || trimmed(source.first_frame_url) || (Array.isArray(source.media) && source.media.length) ) return "image_to_video"; } return "text_to_video"; } function convert(ctx) { const req = ctx.requestBody; const upstreamModel = ctx.upstreamModel || req.model; const profile = modelProfile(upstreamModel); const metadata = objectValue(req.metadata, "metadata"); if (metadata.model !== undefined && metadata.model !== upstreamModel) throw new Error("can't change model with metadata"); const input = {}; for (const key of ["prompt", "negative_prompt", "img_url", "image_url", "first_frame_url", "last_frame_url", "audio_url", "template", "media"]) { if (req[key] !== undefined) input[key] = req[key]; } const image = firstImage(req); if (image && input.img_url === undefined) input.img_url = image; Object.assign(input, objectValue(metadata.input, "metadata.input")); const nativeParameters = objectValue(metadata.parameters, "metadata.parameters"); const parameters = profile.kind === "speech" ? {} : { prompt_extend: true }; for (const key of ["resolution", "ratio", "prompt_extend", "watermark", "audio", "seed", "shot_type"]) { if (req[key] !== undefined) parameters[key] = req[key]; } Object.assign(parameters, nativeParameters); for (const key of ["prompt_extend", "watermark", "audio"]) { if (parameters[key] != null && typeof parameters[key] !== "boolean") throw new Error(key + " must be a boolean"); } if (parameters.seed != null && (!Number.isInteger(parameters.seed) || parameters.seed < (profile.kind === "all" ? -1 : 0) || parameters.seed > 2147483647)) throw new Error("seed must be an integer between " + (profile.kind === "all" ? -1 : 0) + " and 2147483647"); const sizeValue = nativeParameters.size ?? req.size; const size = sizeValue == null || sizeValue === "" ? {} : videoSize(sizeValue); let resolution = normalizeResolution(parameters.resolution ?? size.resolution ?? profile.defaultResolution); let ratio = parameters.ratio ?? size.ratio ?? (profile.kind === "all" ? "adaptive" : "16:9"); // In the legacy T2V protocol the explicit native size is authoritative. if (profile.kind === "size" && nativeParameters.size != null && nativeParameters.size !== "") { resolution = size.resolution; ratio = size.ratio ?? parameters.ratio ?? "16:9"; } if (!profile.resolutions.includes(resolution)) throw new Error(upstreamModel + " resolution must be one of " + profile.resolutions.join(", ")); if (profile.kind === "size" || profile.kind === "t2v" || profile.kind === "all") { const ratios = ["16:9", "9:16", "1:1", "4:3", "3:4"]; if (profile.kind === "all") ratios.push("adaptive"); if (!ratios.includes(ratio)) throw new Error(upstreamModel + " ratio must be one of " + ratios.join(", ")); } if (profile.kind === "size") { const pixelSize = LEGACY_SIZES[resolution][ratio]; if (!pixelSize) throw new Error("unsupported ratio for " + resolution + ": " + ratio); parameters.size = pixelSize; delete parameters.resolution; delete parameters.ratio; } else { parameters.resolution = resolution; delete parameters.size; if (profile.kind === "t2v" || profile.kind === "all") parameters.ratio = ratio; else delete parameters.ratio; // Image-derived aspect ratios are not configurable. } const rawDuration = nativeParameters.duration ?? (req.auto_duration === true ? -1 : (req.duration ?? req.seconds)); if (rawDuration != null && ((typeof rawDuration !== "number" && typeof rawDuration !== "string") || String(rawDuration).trim() === "")) throw new Error("duration must be a number"); const duration = rawDuration == null ? 5 : Number(rawDuration); if (duration === -1 && profile.kind !== "all") throw new Error("duration -1 (smart duration) is only supported by wan3.0 models"); if (profile.kind === "speech") { if (rawDuration != null && (!Number.isFinite(duration) || duration <= 0 || duration >= 20)) throw new Error("wan2.2-s2v duration must be positive and less than 20 seconds; output follows the audio"); delete parameters.duration; } else { if (profile.durations) { if (!profile.durations.includes(duration)) throw new Error(upstreamModel + " duration must be one of " + profile.durations.join(", ")); } else if (!(profile.kind === "all" && duration === -1) && (!Number.isInteger(duration) || duration < 2 || duration > profile.maxDuration)) { throw new Error(upstreamModel + " duration must be " + (profile.kind === "all" ? "-1 or " : "") + "an integer between 2 and " + profile.maxDuration); } parameters.duration = duration; } if (profile.kind === "media" || profile.kind === "all") { if (input.media !== undefined && !Array.isArray(input.media)) throw new Error("input.media must be an array"); if (!input.media || input.media.length === 0) { input.media = []; const first = trimmed(input.first_frame_url) || trimmed(input.img_url) || firstImage(req); const last = trimmed(input.last_frame_url) || secondImage(req); if (first) input.media.push({ type: "first_frame", url: first }); if (last) input.media.push({ type: "last_frame", url: last }); if (trimmed(input.audio_url)) input.media.push({ type: profile.kind === "media" ? "driving_audio" : "reference_audio", url: input.audio_url }); } const counts = {}; const limits = profile.kind === "media" ? { first_frame: 1, last_frame: 1, driving_audio: 1, first_clip: 1 } : { first_frame: 1, last_frame: 1, reference_image: 10, reference_video: 5, reference_audio: 5, file: 1, link: 1 }; for (const media of input.media) { if (!media || !Object.prototype.hasOwnProperty.call(limits, media.type) || !trimmed(media.url)) throw new Error("invalid input.media type or url"); counts[media.type] = (counts[media.type] || 0) + 1; if (counts[media.type] > limits[media.type]) throw new Error("too many input.media entries for " + media.type); } if (profile.kind === "media") { if ((!counts.first_frame && !counts.first_clip) || (counts.first_clip && (counts.first_frame || counts.driving_audio))) throw new Error("wan2.7-i2v requires first_frame or first_clip with a supported media combination"); } else { const frames = counts.first_frame || counts.last_frame; const references = counts.reference_image || counts.reference_video || counts.reference_audio || counts.file || counts.link; if ((frames && references) || (counts.file && counts.link) || (counts.last_frame && !counts.first_frame)) throw new Error("unsupported wan3.0 input.media combination"); if (!trimmed(input.prompt) && !input.media.length) throw new Error("wan3.0-video requires prompt or input.media"); } if (!input.media.length) delete input.media; delete input.img_url; delete input.first_frame_url; delete input.last_frame_url; delete input.audio_url; } else if (profile.kind === "frames") { input.first_frame_url = trimmed(input.first_frame_url) || trimmed(input.img_url); if (!input.first_frame_url) throw new Error(upstreamModel + " requires first_frame_url or image"); if (!input.last_frame_url && secondImage(req)) input.last_frame_url = secondImage(req); delete input.img_url; } else if (profile.kind === "speech") { input.image_url = trimmed(input.image_url) || trimmed(input.img_url); if (!input.image_url || !trimmed(input.audio_url)) throw new Error("wan2.2-s2v requires image_url and audio_url"); delete input.img_url; if (!trimmed(input.prompt)) delete input.prompt; } else if (profile.kind === "image") { if (!trimmed(input.img_url)) throw new Error(upstreamModel + " requires img_url or image"); } else if (!trimmed(input.prompt)) { throw new Error("input is required"); } return { model: upstreamModel, input: input, parameters: parameters }; } function resolutionRatio(body) { const resolution = body.parameters.size ? videoSize(body.parameters.size).resolution : body.parameters.resolution; const ratios = { "wan3.0-video": { "480P": 1, "720P": 2, "1080P": 4 }, "wan3.0-video-prime": { "480P": 1, "720P": 2, "1080P": 4 }, "wan2.6-i2v": { "720P": 1, "1080P": 1 / 0.6 }, "wan2.5-t2v-preview": { "480P": 1, "720P": 2, "1080P": 1 / 0.3 }, "wan2.2-t2v-plus": { "480P": 1, "1080P": 5 }, "wan2.5-i2v-preview": { "480P": 1, "720P": 2, "1080P": 1 / 0.3 }, "wan2.2-i2v-plus": { "480P": 1, "1080P": 5 }, "wan2.2-kf2v-flash": { "480P": 1, "720P": 2, "1080P": 4.8 }, "wan2.2-i2v-flash": { "480P": 1, "720P": 2 }, "wan2.2-s2v": { "480P": 1, "720P": 1.8 }, }; const model = modelKey(body.model); return ratios[model] ? { key: "resolution-" + resolution, value: ratios[model][resolution] } : null; } function responsesInput(req) { const texts = [], images = []; const input = req.input; if (typeof input === "string") texts.push(input); else if (Array.isArray(input)) { for (const item of input) { if (typeof item === "string") { texts.push(item); continue; } if (!item || typeof item !== "object" || Array.isArray(item)) continue; const content = item.content === undefined ? [item] : Array.isArray(item.content) ? item.content : [item.content]; for (const part of content) { if (typeof part === "string") { texts.push(part); continue; } if (!part || typeof part !== "object" || Array.isArray(part)) continue; if (["input_text", "text"].includes(part.type) && typeof part.text === "string") texts.push(part.text); if (["input_image", "image_url"].includes(part.type)) { let image = part.image_url; if (image && typeof image === "object") image = image.url; if (trimmed(image)) images.push(trimmed(image)); } } } } return { prompt: texts .filter(function (text) { return trimmed(text); }) .join("\n"), images: images, }; } function responsesVideoText(ctx) { const artifact = ctx && ctx.artifacts && ctx.artifacts.video; const url = trimmed(artifact && artifact.url); if (!url) throw new Error("video artifact is unavailable"); const escaped = url .replace(/&/g, "&") .replace(/\u0022/g, """) .replace(//g, ">"); return ''; } function responsesOutputText(ctx, task) { const content = imageContent(artifactData(task)); if (!content.length) return responsesVideoText(ctx); const parts = []; let index = 0; for (const part of content) { if (part.image) { const key = "image-" + ++index; const artifact = (ctx.artifacts || {})[key]; if (!artifact || !artifact.url) throw new Error("image artifact is unavailable"); parts.push("![Image " + index + "](<" + artifact.url + ">)"); } else parts.push(part.text); } return parts.join("\n\n"); } export function buildSubmitRequest(ctx) { if (imageModel(ctx)) { const converted = convertImage(ctx); const headers = { Authorization: "Bearer " + ctx.apiKey, "Content-Type": "application/json" }; if (!converted.synchronous) headers["X-DashScope-Async"] = "enable"; // A gateway's native route aggregates the vendor stream itself and answers // with one JSON body, so only DashScope is asked for SSE. const streaming = converted.synchronous && converted.body.parameters.enable_interleave === true && !viaNewAPI(ctx); if (streaming) headers["X-DashScope-Sse"] = "enable"; return { url: apiRoot(ctx) + "/api/v1/services/aigc/" + converted.service, method: "POST", headers: headers, body: converted.body, action: converted.action, responseType: streaming ? "sse" : "json", }; } const body = convert(ctx); const kind = modelProfile(body.model).kind; const service = kind === "frames" || kind === "speech" ? "image2video" : "video-generation"; return { url: apiRoot(ctx) + "/api/v1/services/aigc/" + service + "/video-synthesis", method: "POST", headers: { Authorization: "Bearer " + ctx.apiKey, "Content-Type": "application/json", "X-DashScope-Async": "enable" }, body: body, action: videoAction(body.input), }; } // DashScope sends text/image deltas but cumulative usage. The host applies // generic JSON changes; vendor completion and usage semantics remain here. export function parseSubmitEventDelta(ctx, event, previousState) { const body = JSON.parse(event.data); if (!body || typeof body !== "object" || Array.isArray(body)) throw new Error("invalid DashScope stream event"); if (body.code || event.event === "error") throw new Error((body.code || "upstream_error") + ": " + (body.message || "stream failed")); const state = utils.json.clone(previousState || { choices: [], hasUsage: false }); const changes = []; if (!previousState) changes.push({ op: "set", path: [], value: { output: { choices: [], finished: false } } }); if (body.request_id) changes.push({ op: "set", path: ["request_id"], value: body.request_id }); if (body.usage) { if (!state.hasUsage) { changes.push({ op: "set", path: ["usage"], value: {} }); state.hasUsage = true; } for (const key of Object.keys(body.usage)) changes.push({ op: "set", path: ["usage", key], value: body.usage[key] }); } const choices = (body.output || {}).choices || []; for (let index = 0; index < choices.length; index++) { const choice = choices[index]; if (!choice || !choice.message || !Array.isArray(choice.message.content)) throw new Error("invalid DashScope stream choice"); if (!state.choices[index]) { state.choices.push({ count: 0, lastText: false, finishReason: "" }); changes.push({ op: "append", path: ["output", "choices"], value: { message: { role: "assistant", content: [] } } }); } const control = state.choices[index]; for (const part of choice.message.content) { if (!part || typeof part !== "object") throw new Error("invalid DashScope stream content"); if (typeof part.text === "string" && control.lastText) { changes.push({ op: "appendText", path: ["output", "choices", index, "message", "content", control.count - 1, "text"], value: part.text }); } else { changes.push({ op: "append", path: ["output", "choices", index, "message", "content"], value: part }); control.count++; control.lastText = typeof part.text === "string"; } } if (choice.finish_reason && choice.finish_reason !== "null") { control.finishReason = choice.finish_reason; changes.push({ op: "set", path: ["output", "choices", index, "finish_reason"], value: choice.finish_reason }); } } // Some intermediate frames carry finished:true; only terminal finish reasons // make the complete output ready for parsing and billing. const done = state.choices.length > 0 && state.choices.every(function (choice) { return choice.finishReason === "stop"; }); changes.push({ op: "set", path: ["output", "finished"], value: done }); return { changes: changes, state: state, done: done }; } export function parseSubmitResponse(ctx, resp) { const body = resp.body || {}; if (body.code) throw new Error(body.code + ": " + (body.message || "")); if (imageModel(ctx) && convertImage(ctx).synchronous) { const content = imageContent(body); const images = content.filter(function (part) { return part.image; }); const interleaved = convertImage(ctx).body.parameters.enable_interleave === true; if ((!images.length && !(interleaved && content.length)) || (body.output || {}).finished === false) throw new Error("synchronous image response has no completed output"); // No vendor task exists for a synchronous HTTP response. The host already // allocated a public ID; retain that identity without inventing a pollable ID. const taskId = ctx.publicTaskId || body.request_id; if (!taskId) throw new Error("missing synchronous image request id"); return { taskId: taskId, taskData: body, immediate: { status: "SUCCESS", progress: "100%", url: images.length ? images[0].image : "" } }; } if (!body.output || !body.output.task_id) throw new Error("task_id is empty"); return { taskId: body.output.task_id, taskData: body }; } export function extractUsage(ctx) { if (imageModel(ctx)) { const converted = convertImage(ctx); const estimate = imageEstimate(ctx, converted); if (ctx.usagePurpose === "billing_ratios") return imageRatios(ctx, converted, estimate.image_count); return estimate; } const body = convert(ctx); const kind = modelProfile(body.model).kind; const hasVideo = (body.input.media || []).some(function (media) { return media.type === "reference_video"; }); let seconds = body.parameters.duration; if (kind === "speech") seconds = 20; else if (kind === "all" && (hasVideo || seconds === -1)) seconds = 30; if (ctx.usagePurpose === "billing_ratios") { const ratios = { seconds: seconds }; const resolution = resolutionRatio(body); if (resolution && resolution.value !== undefined) ratios[resolution.key] = resolution.value; return ratios; } const resolution = body.parameters.size ? videoSize(body.parameters.size).resolution : body.parameters.resolution; const facts = { seconds: seconds, resolution: resolution }; if (WAN_AUDIO_VIDEO_MODELS.includes(modelKey(body.model))) facts.audio = body.parameters.audio !== false; return facts; } export function extractUsageOnSubmit(ctx, body) { // Legacy ratio pricing uses this hook; task expressions use the same actual // facts through extractUsageOnComplete for both immediate and polled results. if (!imageModel(ctx)) return {}; const converted = convertImage(ctx); if (!converted.synchronous) return {}; const actual = imageUsage(ctx, body || {}); return actual.image_count === undefined ? {} : imageRatios(ctx, converted, actual.image_count); } export function extractUsageOnComplete(task, taskResult, body) { if (imageModel(task)) return imageUsage(task, body || {}); const output = (body && body.output) || {}; const usage = (body && body.usage) || {}; const facts = {}; const model = modelKey(task && (task.upstreamModel || task.model)); if (model === "wan3.0-video" || model === "wan3.0-video-prime" || (!model && usage.input_video_duration != null)) { const inputSeconds = usage.input_video_duration; const outputSeconds = usage.output_video_duration ?? usage.duration; // Partial Wan3 statistics cannot establish the total billable duration. // Keep the reservation when either side is missing. Invalid supplied facts // reach the host validator unchanged, which logs and rejects them. if (inputSeconds != null && outputSeconds != null) { if (typeof inputSeconds !== "number" || !Number.isFinite(inputSeconds) || inputSeconds < 0) facts.seconds = inputSeconds; else if (typeof outputSeconds !== "number" || !Number.isFinite(outputSeconds) || outputSeconds < 0) facts.seconds = outputSeconds; else facts.seconds = inputSeconds + outputSeconds; } } else { const seconds = usage.duration ?? usage.output_video_duration ?? output.duration ?? output.duration_seconds; if (seconds != null) facts.seconds = seconds; } const resolution = usage.SR ?? output.resolution; if (resolution != null) facts.resolution = normalizeResolution(resolution); if (WAN_AUDIO_VIDEO_MODELS.includes(model) && typeof usage.audio === "boolean") facts.audio = usage.audio; return facts; } export function buildQueryRequest(ctx) { return { url: apiRoot(ctx) + "/api/v1/tasks/" + ctx.taskId, method: "GET", headers: { Authorization: "Bearer " + ctx.apiKey } }; } export function parseTaskResult(ctx, body) { const output = body.output || {}; if (output.task_status === "PENDING") return { status: "QUEUED" }; if (output.task_status === "RUNNING") return { status: "IN_PROGRESS" }; if (output.task_status === "SUCCEEDED") { if (imageModel(ctx)) { const content = imageContent(body); const images = content.filter(function (part) { return part.image; }); if (!images.length) { if ( imageModel(ctx) === "image26" && content.some(function (part) { return trimmed(part.text); }) ) return { status: "SUCCESS" }; return { status: "FAILURE", reason: "image task succeeded without any images" }; } return { status: "SUCCESS", url: images[0].image }; } return { status: "SUCCESS", url: videoURL(body) }; } if (["FAILED", "CANCELED", "UNKNOWN"].includes(output.task_status)) { let reason = body.message || ""; if (!reason && output.message) reason = "task failed, code: " + (output.code || "") + " , message: " + output.message; if (!reason) reason = "task failed"; return { status: "FAILURE", reason: reason }; } return { status: "UNKNOWN", reason: "unrecognized status: " + String(output.task_status || "") }; } function artifactData(ctx) { const data = (ctx && ctx.data) || {}; if (data.data && typeof data.data === "object" && data.data.task_id && Object.prototype.hasOwnProperty.call(data.data, "data")) return data.data.data || {}; return data; } function videoURL(body) { const output = (body && body.output) || {}; return trimmed(output.video_url) || trimmed((output.results || {}).video_url); } export function listArtifacts(task) { if (task.status !== "SUCCESS") return []; const body = artifactData(task); const images = imageContent(body).filter(function (part) { return part.image; }); if (images.length) return images.map(function (_, index) { return { key: "image-" + (index + 1), type: "image" }; }); return videoURL(body) ? [{ key: "video", type: "video" }] : []; } export function buildContentRequest(ctx) { let url; if (ctx.artifactKey === "video") url = videoURL(artifactData(ctx)); else { const images = imageContent(artifactData(ctx)).filter(function (part) { return part.image; }); const index = images.findIndex(function (_, index) { return ctx.artifactKey === "image-" + (index + 1); }); url = index >= 0 ? images[index].image : ""; } if (!url) throw new Error("artifact_not_found"); return { url: url, method: ctx.clientRequest.method, credentialless: true }; } export const native = { createImageTask: function (ctx) { if (!ctx.body || ctx.body.kind !== "json" || !ctx.body.value || typeof ctx.body.value !== "object" || Array.isArray(ctx.body.value)) throw new Error("JSON object required"); const req = ctx.body.value; if (req.stream !== undefined && req.stream !== false) throw new Error("native task responses are aggregated JSON; configure upstream streaming in parameters"); const requestBody = { model: req.model, metadata: { input: objectValue(req.input, "input"), parameters: objectValue(req.parameters, "parameters"), upstream_mode: String(ctx.path || "").includes("/multimodal-generation/") ? "sync" : "async", }, }; return { kind: "submit", model: req.model, requestBody: requestBody }; }, imageCreated: function (ctx, task) { return task.data || {}; }, createVideoTask: function (ctx) { if (!ctx.body || ctx.body.kind !== "json" || !ctx.body.value || typeof ctx.body.value !== "object" || Array.isArray(ctx.body.value)) throw new Error("JSON object required"); const req = ctx.body.value, input = objectValue(req.input, "input"), parameters = objectValue(req.parameters, "parameters"); const requestBody = { model: req.model, metadata: { input: input, parameters: parameters }, }; if (input.prompt !== undefined) requestBody.prompt = input.prompt; const image = input.img_url || input.image_url || input.first_frame_url; if (image !== undefined) requestBody.image = image; return { kind: "submit", model: req.model, action: videoAction(requestBody), requestBody: normalizeRequest(requestBody), }; }, taskCreated: function (ctx, task) { const data = task.data || {}; return { request_id: data.request_id || "", output: { task_id: task.task_id, task_status: "PENDING" } }; }, taskStatus: function (ctx, task) { const data = task.data || {}, output = Object.assign({}, data.output || {}, { task_id: task.task_id }); return Object.assign({}, data, { output: output }); }, error: function (ctx, error) { return { code: error.code, message: error.message, request_id: "" }; }, }; export const protocols = { openai_responses: { decodeRequest: function (ctx) { if (!ctx.body || ctx.body.kind !== "json") throw new Error("JSON body required"); const req = ctx.body.value; if (!req || typeof req !== "object" || Array.isArray(req)) throw new Error("request body must be an object"); const model = trimmed(ctx.model); if (!model) throw new Error("model is required"); if (req.input !== undefined && typeof req.input !== "string" && !Array.isArray(req.input)) throw new Error("input must be a string or array"); const input = responsesInput(req); const prompt = input.prompt || trimmed(req.prompt); let requestBody = { model: model, prompt: prompt }; if (trimmed(req.image)) requestBody.image = trimmed(req.image); if (req.images !== undefined && !Array.isArray(req.images)) throw new Error("images must be an array"); const images = []; for (const image of req.images || []) if (trimmed(image) && !images.includes(trimmed(image))) images.push(trimmed(image)); for (const image of input.images) if (!images.includes(image)) images.push(image); if (images.length) requestBody.images = images; if (trimmed(req.input_reference)) requestBody.input_reference = trimmed(req.input_reference); for (const key of [ "size", "resolution", "ratio", "duration", "seconds", "auto_duration", "prompt_extend", "watermark", "audio", "seed", "shot_type", "negative_prompt", "img_url", "image_url", "first_frame_url", "last_frame_url", "audio_url", "template", "media", ]) { if (Object.prototype.hasOwnProperty.call(req, key)) requestBody[key] = req[key]; } if (Object.prototype.hasOwnProperty.call(req, "metadata")) requestBody.metadata = req.metadata; if (imageModel({ upstreamModel: ctx.upstreamModel, model: model })) { for (const key of ["n", "enable_interleave", "max_images", "enable_sequential", "thinking_mode", "bbox_list", "color_palette"]) { if (Object.prototype.hasOwnProperty.call(req, key)) requestBody[key] = req[key]; } // The model may be a mapped alias; final validation runs after channel selection. return { kind: "submit", model: model, action: firstImage(requestBody) ? "image_to_image" : "text_to_image", requestBody: requestBody }; } requestBody = normalizeRequest(requestBody); const action = videoAction(requestBody); const profile = WAN_MODELS[modelKey(ctx.upstreamModel || model)]; const nativeInput = objectValue((requestBody.metadata || {}).input, "metadata.input"); if (!trimmed(nativeInput.prompt ?? prompt) && (action === "text_to_video" || (profile && (profile.kind === "size" || profile.kind === "t2v")))) throw new Error("input is required"); return { kind: "submit", model: model, action: action, requestBody: requestBody }; }, renderEvents: function (ctx, task, previousState) { const status = String(task.status || "UNKNOWN").toUpperCase(); const value = Number(String(task.progress || "").replace("%", "")); const progress = Number.isFinite(value) && value >= 0 && value <= 100 ? value : null; const state = { status: status, progress: progress }; if (status === "SUCCESS") { const text = responsesOutputText(ctx, task); const events = previousState && previousState.status === status ? [] : text ? [{ type: "output", data: text }] : []; return { events: events, state: state, done: true }; } if (status === "FAILURE") { return { events: [{ type: "error", code: "task_failed", message: "task failed" }], state: state, done: true }; } if (previousState && previousState.status === status && previousState.progress === progress) { return { events: [], state: state, done: false }; } const event = { type: "progress", message: status.toLowerCase() }; if (progress !== null) event.progress = progress; return { events: [event], state: state, done: false }; }, renderFinal: function (ctx, task) { return { output: [ { type: "message", status: "completed", role: "assistant", content: [ { type: "output_text", text: responsesOutputText(ctx, task), annotations: [], logprobs: [], }, ], }, ], metadata: { vendor: "ali" }, }; }, }, // OpenAI Images API. The host pins ctx.model, waits for the task to become // terminal and honors response_format itself; the plugin maps the request // onto the DashScope image services and renders data[] from the result. openai_image: { decodeRequest: function (ctx) { const model = trimmed(ctx.model); if (!model) throw new Error("model is required"); let req = {}; const uploads = []; if (ctx.body && ctx.body.kind === "json") { req = ctx.body.value; if (!req || typeof req !== "object" || Array.isArray(req)) throw new Error("request body must be an object"); } else if (ctx.body && ctx.body.kind === "multipart") { const fields = ctx.body.fields || {}; for (const name of Object.keys(fields)) { if (fields[name].length > 1) throw new Error(name + " must be provided once"); req[name] = fields[name][0]; } for (const key of ["n", "seed", "max_images", "upscale_factor", "strength", "top_scale", "bottom_scale", "left_scale", "right_scale"]) { if (req[key] !== undefined) req[key] = Number(req[key]); } for (const key of ["prompt_extend", "watermark", "enable_thinking", "enable_interleave", "enable_sequential", "thinking_mode", "is_sketch"]) { if (req[key] === undefined) continue; if (req[key] !== "true" && req[key] !== "false") throw new Error(key + " must be true or false"); req[key] = req[key] === "true"; } for (const key of ["parameters", "input"]) { if (req[key] === undefined) continue; try { req[key] = JSON.parse(req[key]); } catch (e) { throw new Error(key + " must be a JSON object string"); } } for (const file of ctx.body.files || []) { const upload = { __fileRef: file.ref, encoding: "dataUrl", mimeType: trimmed(file.mimeType) || "image/png", maxBytes: MAX_INPUT_IMAGE_BYTES }; // A mask upload is not a reference image; only wanx2.1-imageedit reads it. if (file.field === "mask") req.mask = upload; else if (/^image(\[\d*\])?$/.test(file.field)) uploads.push(upload); } } else throw new Error("JSON or multipart body required"); if (req.stream !== undefined && req.stream !== false && req.stream !== "false") throw new Error("stream is not supported; the complete image response is returned once all images are generated"); if (req.response_format !== undefined && req.response_format !== "url" && req.response_format !== "b64_json") throw new Error("response_format must be url or b64_json"); const requestBody = { model: model, prompt: typeof req.prompt === "string" ? req.prompt : "" }; if (!trimmed(requestBody.prompt) && !objectValue(req.input, "input").messages && !objectValue(req.input, "input").prompt) throw new Error("prompt is required"); requestBody.n = req.n === undefined || req.n === null ? 1 : req.n; if (!Number.isInteger(requestBody.n) || requestBody.n < 1) throw new Error("n must be a positive integer"); const images = []; for (const image of [].concat(req.image === undefined ? [] : req.image, req.images === undefined ? [] : req.images)) { if (isImageInput(image) || (typeof image === "string" && trimmed(image))) images.push(typeof image === "string" ? trimmed(image) : image); else throw new Error("image must be an HTTP URL or Base64 data URL"); } for (const upload of uploads) images.push(upload); if (ctx.operation === "edit" && !images.length) throw new Error("image is required"); if (images.length) requestBody.images = images; for (const key of [ "size", "negative_prompt", "prompt_extend", "prompt_extend_mode", "enable_thinking", "watermark", "seed", "enable_interleave", "max_images", "enable_sequential", "thinking_mode", "bbox_list", "color_palette", "function", "mask", "strength", "top_scale", "bottom_scale", "left_scale", "right_scale", "upscale_factor", "is_sketch", ]) { if (Object.prototype.hasOwnProperty.call(req, key)) requestBody[key] = req[key]; } // Provider passthrough: the same parameters/input objects the DashScope API accepts. const metadata = {}; if (req.parameters !== undefined) metadata.parameters = objectValue(req.parameters, "parameters"); if (req.input !== undefined) metadata.input = objectValue(req.input, "input"); if (Object.keys(metadata).length) requestBody.metadata = metadata; // The model may be a mapped alias; final validation runs after channel selection. return { kind: "submit", model: model, action: images.length ? "image_to_image" : "text_to_image", requestBody: requestBody }; }, render: function (ctx, task) { const data = []; for (const part of imageContent(artifactData(task))) { if (part.image) data.push(imageDatum(part.image)); } return { created: task.created_at, data: data, metadata: artifactData(task) }; }, }, openai_video: { decodeRequest: function (ctx) { let req; if (ctx.body && ctx.body.kind === "json") req = ctx.body.value; else if (ctx.body && ctx.body.kind === "multipart") { if ((ctx.body.files || []).length) throw new Error("Alibaba requires image references to be URLs"); const first = function (name) { const values = (ctx.body.fields || {})[name] || []; if (values.length > 1) throw new Error(name + " must be provided once"); return values[0]; }; req = {}; const fields = ctx.body.fields || {}; for (const name of Object.keys(fields)) { if (name === "images") req.images = fields[name] || []; else req[name] = first(name); } if (req.metadata !== undefined) { let parsed; try { parsed = JSON.parse(req.metadata); } catch (e) { throw new Error("metadata must be a JSON object string"); } if (!parsed || typeof parsed !== "object" || Array.isArray(parsed)) throw new Error("metadata must be a JSON object string"); req.metadata = parsed; } if (req.seconds !== undefined) req.seconds = Number(req.seconds); else if (req.duration !== undefined) req.seconds = Number(req.duration); if (req.duration !== undefined) req.duration = Number(req.duration); if (req.seed !== undefined) req.seed = Number(req.seed); for (const key of ["prompt_extend", "watermark", "audio", "auto_duration"]) { if (req[key] === undefined) continue; if (req[key] !== "true" && req[key] !== "false") throw new Error(key + " must be true or false"); req[key] = req[key] === "true"; } } else throw new Error("JSON or multipart body required"); if (!req || typeof req !== "object" || Array.isArray(req)) throw new Error("request body must be an object"); if (req.images !== undefined && !Array.isArray(req.images)) throw new Error("images must be an array"); const requestBody = normalizeRequest(Object.assign({}, req, { model: ctx.model })); return { kind: "submit", model: ctx.model, action: videoAction(requestBody), requestBody: requestBody, }; }, render: function (ctx, task) { const data = task.data || {}, outputData = data.output || {}; const statuses = { PENDING: "queued", RUNNING: "in_progress", SUCCEEDED: "completed", FAILED: "failed", CANCELED: "failed", UNKNOWN: "failed", }; const output = { id: task.task_id, object: "video", model: task.properties ? task.properties.origin_model_name || "" : "", status: statuses[outputData.task_status] || "unknown", progress: Number(String(task.progress || "0").replace("%", "")), created_at: task.created_at, completed_at: task.updated_at, }; if (data.code) output.error = { code: data.code, message: data.message || "" }; else if (outputData.code) output.error = { code: outputData.code, message: outputData.message || "" }; return output; }, }, };