Skip to content

Wan 3.0 Video

POST /v1/videos

Standard Wan 3.0: 2–30-second generation (or model-decided duration), native synchronized audio, and first/last-frame, reference-image, reference-video, and reference-audio inputs.

  • Resolutions: 480P, 720P, 1080P; unspecified bills at 1080P
  • Duration: 2–30 s, or -1 (model-decided; bills actual seconds)
  • Reference media: images (≤10), videos (≤5, total ≤15 s), audio (≤5, total ≤15 s); first/last frames supported; frames and reference media are mutually exclusive

See the family page for the shared request shape, media limits, and billing rules: Wan 3.0 (DashScope).

API fieldTypeRequiredValues / limit
modelstringYestk-wan3.0-video
promptstringYesNonempty; at most 6,000 characters
secondsintegerNo2–30, or -1 for model-decided duration; default 5
metadata.resolutionstringNo480P, 720P, 1080P; default 1080P
metadata.ratiostringNoadaptive (default), 16:9, 9:16, 1:1, 4:3, 3:4, 21:9
metadata.mediaarrayNo{type, url} entries
metadata.audiobooleanNoGenerated audio track; default true
metadata.watermarkbooleanNoDefault false
metadata.seedintegerNoRandom seed
metadata.prompt_extendbooleanNoModel rewrites the prompt; default true
Terminal window
curl https://api.apimart.cc/v1/videos \
-H "Authorization: Bearer $HIPIX_API_KEY" \
-H "Content-Type: application/json" \
--data-raw '{"model": "tk-wan3.0-video", "prompt": "A cat running on a moonlit rooftop", "seconds": 5, "metadata": {"resolution": "720P", "ratio": "16:9", "audio": true}}'
  • Billed per second at the output resolution rate: 480P, 720P, or 1080P (see the model square).
  • Reference-video input seconds are billed at the same rate: billed seconds = output seconds + reference-video seconds. First/last frames, reference images, and reference audio are not billed.
  • An unspecified metadata.resolution bills at the 1080P rate; always pass it explicitly.
  • Failed tasks are not billed.