Skip to content

PixVerse Video and Image Capabilities ​

Send a POST request to https://modelfold.ai/v1/jobs/createTask with Authorization: Bearer $MODELFOLD_API_KEY and Content-Type: application/json. The request body is { "model": "model ID", "input": { ... } }. Optionally set callBackUrl at the top level, or retrieve results using the returned task ID.

Video generation ​

Use pixverse-<version>/<mode> as the model ID. Versions include c1, v5.6, v5.5, and v5; V5.5 does not offer first-and-last-frame generation. Check the model's API tab for supported durations, resolutions, and audio options.

ModeInput
text-to-videoprompt, duration, quality, and aspect_ratio
image-to-videoimage_url, prompt, duration, and quality
transitionfirst_frame_image_url, last_frame_image_url, prompt, duration, and quality
reference-to-videoimage_references (objects with url and optional ref_name and type), prompt, duration, quality, and aspect_ratio

Media processing and templates ​

Use URLs reachable by the API for media inputs. Submit requests through the API when the web Playground does not support uploading the required media.

Model IDMain inputs
pixverse-multi-transitionOrdered image_urls and segment_durations; supply one more image than segment durations. Choose quality and optionally motion_mode
pixverse-extendvideo_url, prompt, duration (seconds to add), and quality. The compatibility endpoint pixverse-v6/extend takes taskId instead of video_url; the task must have completed with a video and must belong to the same API key.
pixverse-expandvideo_url, quality, and aspect_ratio
pixverse-upscalevideo_url
pixverse-restylevideo_url and exactly one of restyle_id or prompt
pixverse-lipsyncvideo_url and either audio_url or both tts_text and speaker_id
pixverse-avatarimage_url, quality, and either audio_url or both tts_text and speaker_id
pixverse-select-maskvideo_url and optional keyframe_id. The returned mask_ref is scoped to your account and expires after 30 minutes
pixverse-swapThe returned mask_ref, replacement image_url, and quality. Do not resubmit the video
pixverse-mimicMotion reference video_url, target image_url, and quality
pixverse-sound-effectvideo_url and optional prompt and original_sound_switch. Returns video with audio
pixverse-image-templatetemplate_id and ordered image_urls as required by the template. Returns images
pixverse-vibemvaudio_url, quality, and aspect_ratio; optionally add image_urls as references
json
{
  "model": "pixverse-c1/text-to-video",
  "input": {
    "prompt": "Mist rises through a pine forest in early morning light as the camera moves forward",
    "duration": 5,
    "quality": "720p",
    "aspect_ratio": "16:9"
  }
}

ModelFold — one API for all the best AI models.