> ## Documentation Index > Fetch the complete documentation index at: https://evolink.ai/docs/llms.txt > Use this file to discover all available pages before exploring further. # Wan3.0 Reference-to-Video > - Reference images / reference videos / reference audio / files / web pages serve as material; the model interprets the intent automatically and generates the video - Refer to the material in the prompt as "Image 1", "Video 1" and "Audio 1"; **the three material types are counted independently** - `prompt` and reference material: **at least one of the two must be supplied** - **First and last frames are not accepted** (`image_start` / `image_end`): reference material and first-and-last frames are mutually exclusive. To strictly control where the picture starts and ends, use [Wan3.0 Image-to-Video](/en/api-manual/video-series/wan3.0/wan3.0-image-to-video) instead - **🔴 Reference videos count toward billing**, reference images / reference audio / files / web pages are not billed - Asynchronous processing mode, use the returned task ID to [query status](/en/api-manual/task-management/get-task-detail) - Generated video links are valid for 24 hours, please save them promptly ## Billing - **Billing formula**: billed duration = input video duration + output video duration, billed per second - **Resolution multiplier**: `480p` = 1x (baseline), `720p` = 2x, `1080p` = 4x - Reference images, reference audio, reference files and web page links are **not billed** - When `duration` is `-1` (smart duration), credits are pre-authorized at the `30`-second cap; once the task succeeds it is settled against the actual output duration and the difference is refunded automatically - Turning the `audio` track on or off **costs the same** - **Failed tasks are not billed**, the frozen credits are refunded in full ## OpenAPI ````yaml en/api-manual/video-series/wan3.0/wan3.0-reference-video.json POST /v1/videos/generations openapi: 3.1.0 info: title: Wan3.0 Reference-to-Video API description: >- Wan 3.0 reference-to-video, supporting mixed multimodal input: reference images, reference videos, reference audio, files and web pages license: name: MIT version: 1.0.0 servers: - url: https://api.evolink.ai description: Production security: - bearerAuth: [] tags: - name: Video Generation description: AI video generation endpoints paths: /v1/videos/generations: post: tags: - Video Generation summary: Wan3.0 Reference-to-Video description: >- - Reference images / reference videos / reference audio / files / web pages serve as material; the model interprets the intent automatically and generates the video - Refer to the material in the prompt as "Image 1", "Video 1" and "Audio 1"; **the three material types are counted independently** - `prompt` and reference material: **at least one of the two must be supplied** - **First and last frames are not accepted** (`image_start` / `image_end`): reference material and first-and-last frames are mutually exclusive. To strictly control where the picture starts and ends, use [Wan3.0 Image-to-Video](/en/api-manual/video-series/wan3.0/wan3.0-image-to-video) instead - **🔴 Reference videos count toward billing**, reference images / reference audio / files / web pages are not billed - Asynchronous processing mode, use the returned task ID to [query status](/en/api-manual/task-management/get-task-detail) - Generated video links are valid for 24 hours, please save them promptly ## Billing - **Billing formula**: billed duration = input video duration + output video duration, billed per second - **Resolution multiplier**: `480p` = 1x (baseline), `720p` = 2x, `1080p` = 4x - Reference images, reference audio, reference files and web page links are **not billed** - When `duration` is `-1` (smart duration), credits are pre-authorized at the `30`-second cap; once the task succeeds it is settled against the actual output duration and the difference is refunded automatically - Turning the `audio` track on or off **costs the same** - **Failed tasks are not billed**, the frozen credits are refunded in full operationId: createWan30ReferenceVideo requestBody: required: true content: application/json: schema: $ref: '#/components/schemas/VideoGenerationRequest' examples: multi_modal: summary: Multi-material reference (images + video) value: model: wan3.0-reference-video prompt: >- The character in Video 1 holds Image 3 and plays a soft country folk tune on the chair in Image 4, saying: “What lovely sunshine today.” image_urls: - https://example.com/role.jpg - https://example.com/object.png - https://example.com/guitar.png - https://example.com/chair.png video_urls: - https://example.com/ref_role.mp4 duration: 10 quality: 720p aspect_ratio: adaptive with_audio: summary: With reference audio value: model: wan3.0-reference-video prompt: >- The character in Image 1 dances to the rhythm of Audio 1, with the camera orbiting around them. image_urls: - https://example.com/dancer.jpg audio_urls: - https://example.com/beat.mp3 duration: 10 from_file: summary: Reference file to video (deck to promo) value: model: wan3.0-reference-video prompt: >- Turn this product deck into a high-end product ad, minimalist, futuristic and upmarket overall, with restrained lighting. model_params: file_url: https://example.com/product-deck.pptx duration: 10 quality: 1080p aspect_ratio: '16:9' from_link: summary: Reference web page to video value: model: wan3.0-reference-video prompt: >- Turn the key points of this article into a short infographic-style video. model_params: link_url: https://example.com/article/12345 duration: 15 responses: '200': description: Video generation task created successfully content: application/json: schema: $ref: '#/components/schemas/VideoGenerationResponse' '400': description: Invalid request parameters content: application/json: schema: $ref: '#/components/schemas/ErrorResponse' example: error: code: invalid_parameter message: Invalid request parameters type: invalid_request_error '401': description: Unauthenticated, invalid or expired token content: application/json: schema: $ref: '#/components/schemas/ErrorResponse' example: error: code: unauthorized message: Invalid or expired token type: authentication_error '402': description: Insufficient quota, recharge required content: application/json: schema: $ref: '#/components/schemas/ErrorResponse' example: error: code: insufficient_quota message: Insufficient quota. Please top up your account. type: insufficient_quota '403': description: Access denied content: application/json: schema: $ref: '#/components/schemas/ErrorResponse' example: error: code: model_access_denied message: 'Token does not have access to model: wan3.0-reference-video' type: invalid_request_error '429': description: Rate limit exceeded content: application/json: schema: $ref: '#/components/schemas/ErrorResponse' example: error: code: rate_limit_exceeded message: Too many requests, please try again later type: rate_limit_error '500': description: Internal server error content: application/json: schema: $ref: '#/components/schemas/ErrorResponse' example: error: code: internal_error message: Internal server error type: api_error components: schemas: VideoGenerationRequest: type: object required: - model properties: model: type: string description: Model name, fixed to `wan3.0-reference-video` enum: - wan3.0-reference-video example: wan3.0-reference-video default: wan3.0-reference-video prompt: type: string description: >- Text prompt for video generation. Chinese and English are supported, **each Chinese character / letter counts as 1 character**, maximum length `20000` characters; anything beyond that is truncated automatically (no error) **Material reference rules:** - Use "Image 1", "Image 2" to refer to the images at the matching positions in `image_urls` (1-based) - Use "Video 1", "Video 2" to refer to the videos at the matching positions in `video_urls` - Use "Audio 1", "Audio 2" to refer to the audio at the matching positions in `audio_urls` - **Images, videos and audio are counted independently**, so "Image 1" and "Video 1" can both exist - In English prompts write "Image 1", "Video 1", "Audio 1" (capitalised, with a space between the word and the number) example: >- The character in Video 1 holds Image 3 and plays a soft country folk tune on the chair in Image 4, saying: “What lovely sunshine today.” image_urls: type: array items: type: string format: uri maxItems: 10 description: >- Array of reference image URLs, **up to 10 images**. They can supply subjects (people / animals / objects) or scene backgrounds; when a subject is included, each image should contain a **single** character The array order matches "Image 1, Image 2, ..." in the prompt. **Image limits:** - Formats: JPEG, JPG, PNG (transparency not supported), BMP, WEBP - Resolution: width and height in `[240, 8000]` pixels - Aspect ratio: 1:8 ~ 8:1 - File size: up to `20MB` example: - https://example.com/ref1.jpg - https://example.com/ref2.png video_urls: type: array items: type: string format: uri maxItems: 5 description: >- Array of reference video URLs, **up to 5 clips**. The array order matches "Video 1, Video 2, ..." in the prompt **Duration limits:** - `1 ~ 15` seconds per clip - **15 seconds in total at most** - Total reference video duration + output video duration **must not exceed 30 seconds** **🔴 Reference videos count toward billing**: billed duration = input video duration + output video duration. Reference images, reference audio, files and web pages are not billed **Video limits:** - Formats: mp4, mov - Resolution: width and height in `[240, 4096]` pixels - Aspect ratio: 1:8 ~ 8:1 - Size per file: up to `100MB` example: - https://example.com/ref_video.mp4 audio_urls: type: array items: type: string format: uri maxItems: 5 description: >- Array of reference audio URLs, **up to 5 clips**. The array order matches "Audio 1, Audio 2, ..." in the prompt **⚠️ Different semantics from Wan2.7:** in Wan2.7, `audio_urls` carried a **voice** bound to a specific reference image / reference video; in Wan3.0 reference audio is **standalone material** referred to directly from the prompt as "Audio 1", and the `model_params.voice_bindings` voice-binding protocol is **not supported**. Rewrite accordingly when migrating from Wan2.7. **Duration limits:** - `1 ~ 15` seconds per clip, 15 seconds in total at most - Reference audio is **not billed** **Audio limits:** - Formats: wav, mp3 - File size: up to `15MB` example: - https://example.com/ref_audio.mp3 model_params: type: object description: >- Model extension parameters. `file_url` and `link_url` are **mutually exclusive**, pick one of the two properties: file_url: type: string format: uri description: >- Reference file URL, **at most 1**. The model interprets the file's content automatically and generates the video from it (for example turning a product deck into a promo video) **Mutually exclusive with `link_url`**, the two cannot be supplied together **File limits:** - Formats: docx, doc, xlsx, xls, pptx, ppt, pdf, txt, key, pages, numbers, md - File size: up to `100MB` - Pages: no more than 50 example: https://example.com/product-deck.pptx link_url: type: string format: uri description: >- Reference web page link, **at most 1**. The model fetches and interprets the page content automatically, then generates the video **Mutually exclusive with `file_url`**, the two cannot be supplied together **Limitation:** only public pages that **require no login** can be parsed (news articles, blog posts, newsletter articles, etc.) example: https://example.com/article/12345 duration: type: integer description: >- Duration of the generated video in seconds, defaults to `5` **Accepted values:** - Any integer between `2` and `30` - `-1`: smart duration, the model decides the output length from the prompt and the input material **How smart duration is billed:** the output length cannot be known at submission time, so credits are pre-authorized at the `30`-second **cap**; once the task succeeds it is settled against the actual output duration and the excess hold is released automatically. If your balance cannot cover the capped hold, pass an explicit number of seconds instead. **Extra constraint when reference videos are supplied:** - Total reference video duration + output video duration **must not exceed 30 seconds** - Example: with 10 seconds of reference video in total, `duration` can be at most 20 default: 5 example: 5 quality: type: string description: >- Video resolution, defaults to `720p` **Options:** - `480p`: lower definition, lowest price (billing baseline) - `720p`: standard definition, this is the default, 2x the price of `480p` - `1080p`: high definition, 4x the price of `480p` enum: - 480p - 720p - 1080p default: 720p example: 720p aspect_ratio: type: string description: >- Video aspect ratio, defaults to `adaptive` **Options:** - `adaptive`: adaptive, the model recommends a suitable aspect ratio based on the input material's ratio and the intent of the prompt, this is the default - `16:9` (landscape), `9:16` (portrait), `1:1` (square), `4:3`, `3:4` enum: - adaptive - '16:9' - '9:16' - '1:1' - '4:3' - '3:4' default: adaptive example: '16:9' generate_audio: type: boolean description: >- Whether the output video contains an audio track, defaults to `true` **Options:** - `true`: the output video contains sound (voices, sound effects, background music), this is the default - `false`: silent video output **Sound on and sound off cost the same**, there is no extra charge. default: true example: true seed: type: integer description: >- Random seed, used to reproduce generation results, random by default **Notes:** - Range: `0` ~ `2147483647` - Fixing the seed reduces variation when iterating on prompts and improves reproducibility minimum: 0 maximum: 2147483647 example: 42 callback_url: type: string description: >- HTTPS callback URL for task completion **Callback timing:** - Triggered when the task is completed, failed, or cancelled - Sent after billing confirmation is complete **Security restrictions:** - Only HTTPS protocol is supported - Callbacks to private IP addresses are prohibited (127.0.0.1, 10.x.x.x, 172.16-31.x.x, 192.168.x.x, etc.) - URL length must not exceed `2048` characters **Callback mechanism:** - Timeout: `10` seconds - Up to `3` retries after failure (at `1`/`2`/`4` seconds after failure respectively) - Callback response body format is consistent with the task query endpoint response format - A 2xx status code is considered successful; other status codes trigger retries format: uri example: https://your-domain.com/webhooks/video-task-completed VideoGenerationResponse: type: object properties: created: type: integer description: Task creation timestamp example: 1761313744 id: type: string description: Task ID example: task-unified-1774857405-abc123 model: type: string description: Actual model name used example: wan3.0-reference-video object: type: string enum: - video.generation.task description: Specific task type progress: type: integer description: Task progress percentage (0-100) minimum: 0 maximum: 100 example: 0 status: type: string description: Task status enum: - pending - processing - completed - failed example: pending task_info: $ref: '#/components/schemas/VideoTaskInfo' description: Video task details type: type: string enum: - text - image - audio - video description: Task output type example: video usage: $ref: '#/components/schemas/VideoUsage' description: Usage and billing information ErrorResponse: type: object properties: error: type: object properties: code: type: string description: Error code identifier message: type: string description: Error description type: type: string description: Error type VideoTaskInfo: type: object properties: can_cancel: type: boolean description: Whether the task can be cancelled example: true estimated_time: type: integer description: Estimated completion time (seconds) minimum: 0 example: 165 video_duration: type: integer description: Video duration (seconds) example: 8 VideoUsage: type: object description: Usage and billing information properties: billing_rule: type: string description: Billing rule enum: - per_call - per_token - per_second example: per_second credits_reserved: type: number description: Estimated credits consumed minimum: 0 example: 50 user_group: type: string description: User group category example: default securitySchemes: bearerAuth: type: http scheme: bearer description: >- ##All endpoints require Bearer Token authentication## **Get API Key:** Visit the [API Key Management Page](https://evolink.ai/dashboard/keys) to obtain your API Key **Add to request header:** ``` Authorization: Bearer YOUR_API_KEY ``` ````