Curiosity · AI Model

MiniMax Hailuo

Hailuo (海螺) is MiniMax's flagship video-generation product, launched in September 2024. It is a text- and image-to-video diffusion-transformer trained by Shanghai-based MiniMax (the team behind the Abab LLM family), and quickly became a social-media favourite for the quality of its motion, physics, and facial animation. Hailuo supports text-to-video, image-to-video, subject-reference shots, and camera-control prompts, offered via a web app and API.

Model specs

Vendor
MiniMax
Family
Hailuo
Released
2024-09
Context window
1 tokens
Modalities
text, vision, video
Input price
n/a
Output price
n/a
Pricing as of
2026-04-20

Strengths

  • High-quality motion and physics
  • Strong facial animation from reference photos
  • Subject-reference and camera-control features
  • Available globally with English interface

Limitations

  • Closed model — credit-based API and web access
  • Max clip length shorter than Kling / Veo
  • Generation queue can be slow under load
  • No fine-tuning or custom training

Use cases

  • Social short-form creative video
  • Portrait and character animation from a reference image
  • Subject-consistent multi-shot sequences
  • Cinematic camera-motion prompts

Benchmarks

BenchmarkScoreAs of
Artificial Analysis video arenatop tier through 2024 H22024-12
User-preference (LMSYS-style)consistent top-3 with Kling and Veo2024-12

Frequently asked questions

What is MiniMax Hailuo?

Hailuo is MiniMax's video-generation model, launched September 2024. It produces short text-to-video and image-to-video clips with strong motion and facial animation.

How do I access Hailuo?

Via hailuoai.video or the MiniMax API. Usage is credit-based.

What is Hailuo best at?

Fluid motion, realistic physics, character animation from reference photos, and cinematic camera prompts. It is a go-to for short social videos.

Sources

  1. MiniMax Hailuo AI — accessed 2026-04-20
  2. Artificial Analysis — video leaderboard — accessed 2026-04-20