video generation
MiniMax H3 (Hailuo 3)
MiniMax open-weight video model, top open model on Artificial Analysis, with 32 kHz stereo native audio, 4 to 15 s clips up to 2K. Its licence excludes the US, EU, UK and South Korea.
Levelmodel
Classvideo generation
Parameters33B
Contextnot in the notes
Released31 Jul 2026
Opennessopen weights restricted
Runs onmulti-GPU node, consumer GPU
Statusactive
Benchmark Results
independent
Artificial Analysis Text-to-Video leaderboardgenerative media
768p; top open-weightsindependent
1,137
8001,0001,200
Score in Elo; the axis starts at 800, as Elo has no zero. Higher is better.
rank 4, AA Text-to-Video leaderboard
LMArena Text-to-Video Arenagenerative media
independent
1,460 ±9
1,2001,4001,600
Score in Elo; the axis starts at 1,200, as Elo has no zero. Higher is better.
rank 8, LMArena Text-to-Video, updated 21 Sep 2026
Where this comes from
Every score names who ran it: vendor-reported, maintainer-run or independent. A rank is shown with the population it was ranked in (gap rule 3).
Lineage
product ofsuccessor
MiniMax H3 Max is a child (product of)low confidencenotesHailuo 2.3 is the parent (successor)medium confidencenotesWhere this comes from
Parent and child relations from model cards, Hugging Face base_model fields and the genealogy tables of the research notes. The tree shows two generations each way, 4 models. Confidence and evidence are the notes' own.
Runtimes
Identity
- Developer
- MiniMax
- Family
- Hailuo / MiniMax video
- Released
- 31 Jul 2026 sourcessecondary report
- Weights published
- 3 Aug 2026
- Record checked
- 10 Oct 2026
Where this comes from
Developer and dates from the research notes. Release date is the vendor's public announcement; weights date is when open weights were published.
Architecture and Size
- Architecture
- diffusion Single-stream DiT; Qwen3-VL-32B text features; 16x/4x VAE; 3 stages (Context-IR, Base 768p, Regenerate-2K)
- Total parameters
- 33B sources
- Active parameters
- –
- Checkpoint parameters
- –
- Note
- 33B Base, dense
- Context length
- not in the notes
- Input
- text
- Output
- video, audio
Where this comes from
Parameter counts as the developer states them. Where the checkpoint holds a different total, both are shown (register A16).
Openness and Access
- Openness
- open weights restricted
- Licence
- MiniMax H3 Community Licence Agreement sources
- Access routes
- open weights download, API
Where this comes from
The licence file governs over registry tags and aggregator labels.
Deployment
Where this comes from
Deployment tiers and runtimes as named in the notes. Footprint is the notes' figure, not a measurement by AURION.
Indonesia
- Indonesian support
- not stated
- Regional languages
- not in the notes
- Jakarta hosting
- unknown
- Languages
- not in the notes
Where this comes from
Indonesia fields are carried on every model; unknown is an allowed value.
Industry Use Cases
No industry use case links this model.
Links
DeveloperMiniMaxstartup, China
MiniMaxstartup, ChinaModelMiniMax H3 (Hailuo 3)MiniMax, video generation
MiniMax H3 (Hailuo 3)MiniMax, video generationUse Casesnone recorded
Who Runs Itnone recorded
Models 3
Organisations 1
Sources
- MiniMax Ships H3 Weights , With the US and EU Excluded rits.shanghai.nyu.edu https://rits.shanghai.nyu.edu/ai/minimax-ships-h3-weights-with-the-us-and-eu-excluded/
- AA-Video-T2V v2.0 Leaderboard - Top AI Video Models | Artificial Analysis artificialanalysis.ai https://artificialanalysis.ai/video/leaderboard/text-to-video
- Text-to-Video Arena: Best AI Video Generators arena.ai https://arena.ai/leaderboard/text-to-video
- Text-to-Video Models – Hugging Face huggingface.co https://huggingface.co/models?pipeline_tag=text-to-video&sort=trending
- MiniMax Group - Wikipedia en.wikipedia.org https://en.wikipedia.org/wiki/MiniMax_(company)
To cite this page: Project AURION, AI Use-Cases Research & Innovation, "MiniMax H3 (Hailuo 3)", https://aurion.semai.id/models/minimax-h3/, records of October 2026.