↓
Rivya AI Docs
Model API guides

Wan 2.2 A14B Turbo API reference

Check Wan 2.2 A14B Turbo availability, inputs, parameters, file requirements, credit pricing, and request examples for Rivya Public API v1. Model ID: wan-2-2-a14b-turbo.

Available via APIText or URL ready; reference modes use Files APIVideo
API model ID

wan-2-2-a14b-turbo

Inputs

text, file

Files API

Required for reference modes

Credit pricing

Credit rate: 12–80 per second of output The final cost is rounded up to whole credits.

Capability

Video

Billing

FIXED

Prompt limit

5000 characters

Request fields

Set model at the top level. Put model-specific controls and reference inputs in params, using the format documented for this model.

KeyTypeRequiredDefaultDescription
modelstringYeswan-2-2-a14b-turboSet model to wan-2-2-a14b-turbo.
promptstringYes-Maximum 5000 characters for this model.
paramsobjectNo-The model-specific params object. Use the documented keys and reference-file format for this model.
client_request_idstringNo-Optional ID for tracing this request in your system. It does not replace Idempotency-Key.

Model parameters

KeyTypeRequiredDefaultRangeOptionsDescription
resolutionselectNo720p-480p, 580p, 720pText and image input support 480p or 720p (40 or 80 credits for 5 seconds). Image plus audio also supports 580p; its 480p / 580p / 720p rates are 12 / 18 / 24 credits per output second.
aspect_ratioselectNo16:9-16:9, 9:16This stays available only in text-to-video because the current public image and image-plus-audio-driven endpoints do not expose a stable `aspect_ratio` field.
enable_prompt_expansionselectNofalse-false, trueTurn this on when you want the upstream service to expand a rough text prompt before generation. It stays limited to text and image runs.
seednumberNo-min 0 / max 2147483647-Keep the prompt, reference inputs, and settings unchanged when reusing a seed. Allowed seed values and defaults depend on the model, and identical results are not guaranteed.
accelerationselectNonone-none, regularA lighter acceleration control that remains limited to the public text and image modes.
num_framesnumberNo80min 40 / max 120 / step 4-Image-plus-audio-driven only. Valid values are 40 to 120, and the number must be divisible by 4. A typical starting point is 80.
frames_per_secondnumberNo16min 4 / max 60 / step 1-Image-plus-audio-driven only. Valid values are 4 to 60. Higher FPS usually feels smoother, but also raises generation pressure.
negative_prompttextNo---Image-plus-audio-driven only. Use it to describe motion failures, unwanted artifacts, or visual elements you want to avoid.
num_inference_stepsnumberNo27min 2 / max 40 / step 1-Image-plus-audio-driven only. Valid values are 2 to 40. Higher steps can improve quality, but they also increase runtime.
guidance_scalenumberNo3.5min 1 / max 10 / step 0.1-Image-plus-audio-driven only. Valid values are 1 to 10. Higher values follow the prompt more closely, but can also make motion feel stiffer.
shiftnumberNo5min 1 / max 10 / step 0.1-Image-plus-audio-driven only. Valid values are 1 to 10. It changes the temporal offset feel of the generation, so it is safest to start near the default.
enable_safety_checkerselectNotrue-true, falseImage-plus-audio-driven only. Turn it on to run safety checks before generation, or turn it off when you need less filtering and accept the trade-off.

Upload policy

Upload reference files through Files API first. Keep the original URLs, verified metadata, and tokens. Follow this model's request example: some models use params.referenceMediaItems, while others use different fields.

Reference rule

Reference media: Upload up to 2 assets. Text mode needs none, image-to-video uses 1 image, and the image-plus-audio-driven path uses 1 image plus 1 audio clip.

Max files

2

Accepted file kinds

image, audio

Duration token

When duration verification is required, use duration_token from /api/v1/files as durationToken in the request. Keep any additional signed metadata this model requires.

kindMax sizeMIME types
image10 MBimage/jpeg, image/png, image/webp
audio10 MBaudio/mpeg, audio/mp4, audio/wav, audio/x-wav, audio/aac, audio/ogg, audio/flac, audio/x-ms-wma

Minimal request

{
  "model": "wan-2-2-a14b-turbo",
  "prompt": "A camera slowly circles a product bottle on a studio table.",
  "params": {
    "resolution": "720p",
    "aspect_ratio": "16:9",
    "enable_prompt_expansion": "false",
    "seed": 0,
    "acceleration": "none",
    "num_frames": 80,
    "frames_per_second": 16,
    "negative_prompt": "Optional exclusions or failure modes to avoid.",
    "num_inference_steps": 27,
    "guidance_scale": 3.5,
    "shift": 5,
    "enable_safety_checker": "true"
  }
}

Reference-file request

{
  "model": "wan-2-2-a14b-turbo",
  "prompt": "A camera slowly circles a product bottle on a studio table.",
  "params": {
    "resolution": "720p",
    "aspect_ratio": "16:9",
    "enable_prompt_expansion": "false",
    "seed": 0,
    "acceleration": "none",
    "num_frames": 80,
    "frames_per_second": 16,
    "negative_prompt": "Optional exclusions or failure modes to avoid.",
    "num_inference_steps": 27,
    "guidance_scale": 3.5,
    "shift": 5,
    "enable_safety_checker": "true",
    "referenceMediaItems": [
      {
        "url": "https://cdn.example.com/reference-image.png",
        "kind": "image",
        "name": "reference-image.png",
        "mimeType": "image/png"
      }
    ]
  }
}

Example response

An accepted generation request returns a task ID. Poll the status endpoint until the task succeeds or fails. A status request failure does not mean generation failed.

{
  "id": "task_wan_2_2_a14b_turbo_example",
  "status": "queued",
  "model": "wan-2-2-a14b-turbo",
  "reserved_credits": 80,
  "final_credits": 0,
  "settlement_status": "RESERVED",
  "created_at": "2026-05-11T00:00:00.000Z",
  "updated_at": "2026-05-11T00:00:00.000Z",
  "result": null,
  "error": null
}

Common errors

validation_failed, insufficient_credits, idempotency_conflict, rate_limited, not_found

Some modes require Files API reference uploads.