Video generation

MiniMax H3

Name: MiniMax H3
Author: MiniMax (Hailuo AI)

MiniMax H3 is a single multimodal model that accepts text, images, video and audio in one content array and returns 4 to 15 second video at 768P or 2K with native stereo sound.

A video generation model by MiniMax (Hailuo AI) · read 2026-08-14 · sourcing high

Up to: 15 seconds
Resolution: 2K (768P or 2K; exact pixel dimensions not published)
Aspect ratios: adaptive, 21:9, 16:9, 4:3
Weights: Not released
API: Yes
Commercial use: Not established

Three request shapes share the endpoint: text-to-video, first/last-frame image-to-video, and reference-to-video using up to 9 images, 3 video clips and 3 audio clips. A separate regeneration endpoint upscales an existing 768P H3 result to 2K, and an H3-Context-IR endpoint returns an enriched prompt without generating video.

Commercial use — Not established

The research notes for this model record other terms but do not state a commercial-use position, and no licence page was retrieved, so it is not established here. Check the vendor's own terms.

The licence notes recorded during research were read for a commercial-use position and did not state one. Read the licence before you rely on any of this. Nothing on this page is legal advice, and this site does not tell you a model is cleared for commercial use.

Specification

Vendor figures, read 2026-08-14Read from MiniMax (Hailuo AI)'s own documentation on 2026-08-14. Vendors change these between model versions without announcing it, so check theirs before you commit a shot to a number. Sourcing for this entry is strong: the figures came from pages we retrieved and read. Check the vendor's current documentation

What it takes in: text, image, video, audio
What it makes: video, audio
Maximum duration: 15 seconds
Maximum resolution: 2K (768P or 2K; exact pixel dimensions not published)
Aspect ratios: adaptive, 21:9, 16:9, 4:3, 1:1, 3:4, 9:16
Controls: image-to-video, start-end-frame, reference-character-consistency, lipsync
Weights: Not released
API: Yes
Pricing: Pay-as-you-go per second of output: $0.13/s at 2K, $0.08/s at 768P. Video input is billed by input duration at the same rates. Audio output is free; images are free for the first 5 then $0.04 each. 768P-to-2K regeneration is $0.05/s. H3-Context-IR is $0.90/M input tokens and $3.60/M output tokens.
Released or updated: 2026-07-31

Strengths

One endpoint covers text, first/last frame and mixed image+video+audio reference generation
Native stereo audio generated jointly with picture, no separate pass
Cheap 2K at $0.13/s; MiniMax claims under a third of mainstream 2K pricing
Regeneration path lets you iterate at 768P then finish at 2K for $0.05/s
Documented input limits are generous: 9 images, 3 videos, 3 audio clips, 7000-character prompts

Weaknesses

No 4K tier
Integer-second durations only, 4 to 15
Requires the pay-as-you-go API key type, not the subscription token plan
Advertised as 'open' but no verifiable weights release as of 2026-08-14

Best for

Developers who want one API for reference-driven video, V2V motion transfer and audio-synced output at low per-second cost.

Where the sources disagree

Recorded rather than resolved. Picking the tidier number would hide the fact that the vendor's own pages do not agree.

{'claim': 'Whether the model weights are open', 'a': "MiniMax's docs banner and release notes call H3 'a new-generation open general-purpose multimodal video model'", 'aUrl': 'https://platform.minimax.io/docs/release-notes/models', 'b': "The launch blog says only that MiniMax 'plan[s] to open up the model weights in the coming days, subject to applicable laws and regulations' — i.e. not released at announcement", 'bUrl': 'https://www.minimax.io/blog/minimax-h3'}

What to understand before using it

Explainers naming the mechanisms this model's own documentation mentions.

Character and set consistency: the weakest link names “consistency”
Temporal consistency: flicker, drift and morphing names “consistency”
Open weights versus closed API: renting or owning the model names “weights”
Latent space: working at proxy scale names “token”

MCP servers that front MiniMax (Hailuo AI)

Servers in the media MCP directory whose own description or homepage names this vendor. Absence of a server here is not evidence that none exists: most publish no tool list at all.

Aetherwave Music
Music, image, video and audio generation across top AI providers - one key, one credit pool.
BlitzReels Video Editor Video generation
Create, edit, reframe, caption, organize, generate, and export short-form videos with BlitzReels.
Clipia Image generation
Generate AI images, video, speech, music and presentations from Claude, ChatGPT and Cursor.
Corent Image generation
Generate images and videos via the Corent API, auto model routing with provider fallback.
Creative Claw Image generation
On-brand creative studio for AI agents: images, video, audio, and 3D.
Distribea Image generation
Generate images, video, music and voice from your CLI or AI agent. On-brand AI media toolkit.

See it from the vendor

Shotcraft hosts no sample output. Judging a model means watching its own showreel with your own eyes, and a clip re-hosted here would be a clip we chose.

Nearby in the directory

MiniMax Hailuo 02

MiniMax-Hailuo-02 is the 2025-generation Hailuo model and the only MiniMax V1 model that supports first-and-last-frame generation as well as text-to-v

Video generation

MiniMax Hailuo 2.3 (and 2.3-Fast)

MiniMax-Hailuo-2.3 is the previous-generation Hailuo video model, still served on MiniMax's V1 text-to-video and image-to-video endpoints. Output is 6

Video generation

Adobe Firefly Video Model

The Adobe Firefly Video Model generates video from a text prompt, optionally guided by an image used as the first or final frame. Adobe's Firefly Serv

Video generation

Sources

Read on 2026-08-14. A source that has changed since then will not be reflected here until the next pass.