> ## Documentation Index > Fetch the complete documentation index at: https://docs.shuyou.ai/llms.txt > Use this file to discover all available pages before exploring further. # Create Prediction > Submit asynchronous image, video, or audio prediction tasks via POST /v1/predictions. ## Common request body All capabilities share the following top-level fields. The shape of `input` varies by model. | Field | Type | Required | Description | | --------- | ------ | -------- | ------------------------------------------------------------------------------------------------------------ | | `model` | string | Yes | Model ID as shown on each model page (may follow `/`). | | `input` | object | Yes | Input parameters for the selected model. See sections below. | | `webhook` | string | No | HTTPS URL for server callbacks on task status or result updates (e.g. `https://api.shuyou.ai/api/callback`). | ## Image generation Use for text-to-image, reference-image generation, and similar scenarios. The following fields are inside the `input` object. Text prompt describing the image to generate. Aspect ratio such as `1:1`, `16:9`, or `9:16`. Supported values depend on the selected model. Output resolution tier such as `1K` or `2K`. Supported values depend on the selected model. Reference or input image URL list (string array) for image-to-image or style reference. Pass `[]` when not needed. Number of images to generate. Default `1`. Actual limit depends on model and quota. Output encoding such as `png`, `jpeg`, or `webp`. Uses platform or model default when omitted. Different models may support different subsets of these fields. See each model page for details. Unsupported fields may be ignored or return a parameter error. ```json Image request theme={null} { "model": "gemini-2.5-flash-image", "input": { "prompt": "Generate a ShuYou AI logo", "aspect_ratio": "1:1", "resolution": "1K", "image_urls": [], "num_images": 1, "output_format": "png" }, "webhook": "https://api.shuyou.ai/api/callback" } ``` ## Video generation Use for text-to-video and related scenarios. The following fields are inside the `input` object. Supported resolution tiers, duration options, and other values vary by model. Text prompt describing video content and camera motion. Aspect ratio such as `16:9`, `9:16`, or `1:1`. Supported values depend on the selected model. Output resolution tier such as `720P` or `1080P`. Supported values depend on the selected model. Video duration, usually passed as a string seconds value or model-specific enum (e.g. `"4"` for 4 seconds). Supported values vary by model. If a field is invalid for the current model, it may be ignored or trigger a validation error. Extended fields (such as first-frame image URLs) are documented on each model page. ```json Video request theme={null} { "model": "veo-3.1-lite-generate-preview", "input": { "prompt": "A cinematic shot of a majestic lion in the savannah.", "aspect_ratio": "16:9", "resolution": "720P", "duration": "4" }, "webhook": "https://api.shuyou.ai/api/callback" } ``` ## Audio generation Audio `input` fields (voice, speed, sample rate, etc.) vary by model. Parameter details and examples will be added in a future release. Refer to the corresponding model page or console documentation when integrating. ```json Response theme={null} { "data": { "task_id": "2c4d50261173430290971a2395a3b607", "task_status": "processing" } } ``` Poll task status with [Get a prediction](/en/api-reference/task-management/get-prediction) using `data.task_id`. ## OpenAPI ````yaml en/api-reference/task-management/create-prediction.json POST /v1/predictions openapi: 3.1.0 info: title: Create Prediction version: 1.0.0 servers: - url: https://api.shuyou.ai security: - bearerAuth: [] paths: /v1/predictions: post: tags: - Task Management summary: Create Prediction description: >- Submit an asynchronous prediction task for image, video, or audio generation. All capabilities share the same top-level request shape; the `input` object varies by model. Poll [Get a prediction](/en/api-reference/task-management/get-prediction) using `data.task_id`, or configure top-level `webhook`. operationId: createPrediction requestBody: required: true content: application/json: schema: $ref: '#/components/schemas/PredictionRequest' examples: image: summary: Image generation value: model: gemini-2.5-flash-image input: prompt: Generate a ShuYou AI logo aspect_ratio: '1:1' resolution: 1K image_urls: [] num_images: 1 output_format: png webhook: https://api.shuyou.ai/api/callback video: summary: Video generation value: model: veo-3.1-lite-generate-preview input: prompt: A cinematic shot of a majestic lion in the savannah. aspect_ratio: '16:9' resolution: 720P duration: '4' webhook: https://api.shuyou.ai/api/callback audio: summary: Audio generation (TTS) value: model: qwen3-tts-flash input: prompt: Welcome to ShuYou AI documentation. voice: Cherry language: Auto webhook: https://api.shuyou.ai/backend/api/callback responses: '200': description: Async task created content: application/json: schema: $ref: '#/components/schemas/PredictionTaskResponse' example: data: task_id: 2c4d50261173430290971a2395a3b607 task_status: processing '400': description: Invalid request '401': description: Unauthorized '402': description: Insufficient quota '403': description: Model access denied '429': description: Rate limit exceeded '500': description: Internal server error components: schemas: PredictionRequest: type: object required: - model - input properties: model: type: string description: >- Model ID as shown on each model page (may follow `/`). example: gemini-2.5-flash-image input: oneOf: - $ref: '#/components/schemas/ImagePredictionInput' - $ref: '#/components/schemas/VideoPredictionInput' - $ref: '#/components/schemas/AudioPredictionInput' description: >- Input parameters for the selected model. See model pages for supported fields. webhook: type: string format: uri description: Optional HTTPS callback URL when task status or results update. example: https://api.shuyou.ai/api/callback PredictionTaskResponse: type: object properties: data: type: object properties: task_id: type: string description: Async task ID. Use with Get a prediction or webhook correlation. example: 2c4d50261173430290971a2395a3b607 task_status: type: string description: Current task status, e.g. `processing`. example: processing required: - task_id - task_status required: - data ImagePredictionInput: type: object required: - prompt properties: prompt: type: string description: Text prompt describing the image to generate. example: Generate a ShuYou AI logo aspect_ratio: type: string description: >- Aspect ratio such as `1:1`, `16:9`, or `9:16`. Supported values depend on the model. example: '1:1' resolution: type: string description: >- Output resolution tier such as `1K` or `2K`. Supported values depend on the model. example: 1K image_urls: type: array description: >- Reference or input image URLs for image-to-image or style reference. Pass `[]` when not needed. items: type: string format: uri default: [] num_images: type: integer description: >- Number of images to generate. Default `1`. Actual limit depends on model and quota. default: 1 example: 1 output_format: type: string description: >- Output encoding such as `png`, `jpeg`, or `webp`. Uses platform or model default when omitted. example: png additionalProperties: true VideoPredictionInput: type: object required: - prompt properties: prompt: type: string description: Text prompt describing video content and camera motion. example: A cinematic shot of a majestic lion in the savannah. aspect_ratio: type: string description: >- Aspect ratio such as `16:9`, `9:16`, or `1:1`. Supported values depend on the model. example: '16:9' resolution: type: string description: >- Output resolution tier such as `720P` or `1080P`. Supported values depend on the model. example: 720P duration: type: string description: >- Video duration, usually passed as a string seconds value or model-specific enum (e.g. `"4"` for 4 seconds). example: '4' additionalProperties: true AudioPredictionInput: type: object description: >- Audio input fields (voice, speed, sample rate, etc.) vary by model. See the corresponding model page. additionalProperties: true securitySchemes: bearerAuth: type: http scheme: bearer description: '`Authorization: Bearer YOUR_API_KEY`' ````