Pruna P Video 2 Pro

Pruna's cinematic generation model, built on MiniMax H3: text or first/last-frame images in, 480p or 768p video with generated audio out, 5 to 15 seconds at 24 fps.

Coming Soon
Text To Video

Pruna P Video 2 Pro is coming to Segmind

The playground and serverless API for this model aren't live yet. Here's what we know so far — launch details, pricing, and full API docs will appear on this page.

About

P Video 2 Pro is Pruna AI's cinematic generation model and the final release in the P-Video family. It is built on MiniMax H3 (also called Hailuo 3), tuned by Pruna for a much faster generation pass than the base model's own API.

Text, a first-frame image, or a first and last frame all go into one endpoint. The model returns a finished scene with generated audio — camera that tracks, motion that holds through the shot, and dialogue in a two-shot. It is aimed at film-style beats: brand films, product shots, game worlds and social.

Pruna reports roughly 2 seconds of generation for a 5-second 480p clip on the default speed recipe, about 4.3 seconds at 768p speed, and about 9.1 seconds at 768p quality.

What to expect

  • Text-to-video from a single prompt
  • First-frame image-to-video, with an optional last-frame lock
  • Generated audio in every output — write the sound or dialogue into the prompt
  • Two generation recipes: speed (default) and quality
  • Prompt upsampling at three levels: off, turbo, max, independent of the recipe
  • Clips of 5 to 15 seconds at 480p or 768p, 24 fps, in seven aspect ratios
  • A reference image sets the canvas, so the aspect ratio follows the image when one is supplied
  • An explicit seed for reproducible reruns

Specifications

Base modelMiniMax H3 (Hailuo 3)
Modalitiestext-to-video, first-frame and first+last-frame image-to-video
Resolutions480p, 768p (default 768p)
Generation modespeed (default), quality
Duration5–15 seconds (default 5)
Frame rate24 fps
Audiogenerated with the video; no audio input
Aspect ratios16:9, 9:16, 4:3, 3:4, 3:2, 2:3, 1:1
Prompt upsampleroff, turbo, max (default turbo)
Seedsupported, for reproducible output

What it is not

P Video 2 Pro is a generation model, not an editor, and its limits are deliberate rather than incidental:

  • It does not edit existing footage — that is P Video Edit.
  • It does not do talking-head avatar workflows — that is P Video Avatar.
  • It takes no audio upload. To drive a clip from a track you already have, use P Video 2.
  • 720p, 1080p, 48 fps and 20-second clips belong to P Video 2, not to Pro.
  • Native 4K is not supported.

MiniMax H3 on Hailuo's own platform also lists 2K output, omni references and native editing. Those are not part of this API.

Availability

P Video 2 Pro goes live on Thursday 17 September 2026. The playground and the API go live on this page, and pricing is published at launch.

Sources