MiniMax's next-generation video route. Create 2K videos from text, first/last frames, or image + video + audio references — with 4–15 second durations on SeedDance.
Available on SeedDance platform

MiniMax H3 is MiniMax's next-generation multimodal video model, succeeding the Hailuo video line. It supports three verified routes — text-to-video, first/last-frame image-to-video, and reference-to-video with ordered images, videos, and audio — delivering 2K MP4 output for ads, ecommerce, and short-form storytelling.
Generate cinematic 2K clips with flexible durations from 4 to 15 seconds — suited for storyboards, product clips, and social creatives.
Animate a start frame, land on an end frame, or transition between both while preserving product shape and character identity.
Guide generation with up to 9 images, 3 videos, and 3 audio clips — for character consistency, motion direction, and voice timbre.
Pick text, frame, or reference mode based on your inputs. Each route uses a dedicated model ID with a strict parameter whitelist.
MiniMax H3 upgrades the Hailuo family with higher resolution, richer references, and production-ready async workflows on SeedDance.

Everything you need to generate 2K MiniMax H3 videos from text, frames, or references.
Describe scenes, camera moves, and style in natural language to generate 2K clips with selectable aspect ratios.
Upload a start frame, end frame, or both. Output aspect ratio follows the input image.
Use ordered image, video, and audio references for character, motion, and sound guidance.
Single 2K quality tier for production-ready MP4 delivery.
Choose any integer length from 4 to 15 seconds per clip.
MiniMax H3 generates audio with every clip — no silent-only mode.
Everything you need to know about MiniMax H3 on SeedDance.
Generate 2K videos from text, frames, or multimodal references on SeedDance.