They are separate models with separate endpoints and pricing. MiniMax H3 LoRA outputs at 480p or 768p (40 Credits or 80 Credits per second), bills every reference image at 20 Credits with no free quota, bills reference audio at 20 Credits per track, and adds a 9:21 aspect ratio. MiniMax H3 outputs at 768p or 2K with different rates and reference rules.
我们注意到您可能偏好使用 简体中文. 要切换语言吗？
Il semble que vous préfériez Français. Souhaitez-vous changer de langue ?
Es scheint, Sie bevorzugen Deutsch. Möchten Sie die Sprache wechseln?
Parece que prefieres Español. ¿Te gustaría cambiar de idioma?
Sembra che tu preferisca Italiano. Vuoi cambiare lingua?
優先言語は次のようです 日本語. 言語を切り替えますか？
선호하시는 언어는 다음과 같습니다 한국어. 언어를 변경하시겠습니까?
Похоже, вы предпочитаете Русский. Хотите сменить язык?
Tercih ettiğiniz dil Türkçe. Dili değiştirmek ister misiniz?
Wygląda na to, że wolisz Polski. Czy chcesz zmienić język?
Black Friday: Up to 66% off most AI models.
Explore the deals
MiniMax H3 LoRA
MiniMax H3 LoRA is the style-focused member of the MiniMax H3 video family. Direct clips from text, images, or rich reference sets — up to nine images, three videos, and three audio tracks — and iterate at 480p for just 20 Credits per second.
Create with MiniMax H3 LoRA
Capabilities
A reference-driven video API built for stylized results
40 Credits Drafting Tier
The 480p tier bills at 40 Credits per output second — the lowest rate in the MiniMax H3 family — so style exploration and motion tests stay inexpensive.
Style-Guided Output
Submit reference sets that describe the look you want and keep characters, palettes, and art direction consistent across clips.
Nine-Image References
Reference-to-video accepts up to nine image URLs in one request, enough to communicate a subject, a wardrobe, and a lighting style together.
Video and Audio Guidance
Add up to three reference videos at 480p and up to three audio tracks to guide motion and sound, billed per actual second or per track.
Seven Frame Shapes
Choose from 21:9, 16:9, 4:3, 1:1, 3:4, 9:16, and the ultra-tall 9:21 format for everything from cinema frames to full-height mobile stories.
Three Generation Modes
Use text-to-video for prompt-only clips, image-to-video with one or two frames, or reference-to-video when style consistency matters most.
Model Details
Published MiniMax H3 LoRA API parameters
MiniMax
Provider
3 Modes
Generation
Video
Output
480p / 768p
Resolution
5–15 Whole Seconds
Duration
Up to 9 Images
Reference Input
Use Cases
Where style-guided, low-cost video generation fits
Style Exploration
Cheap Drafts at 480p
Try several visual directions at 40 Credits per second before committing to a final render. A five-second 480p draft costs 200 Credits, so ten style candidates come to 2,000 Credits — then switch to 768p once the direction is settled.
Brand Consistency
Recurring Characters and Looks
Feed up to nine reference images of a character, product, or art style into reference-to-video and keep the same identity across a whole campaign of short clips.
Motion References
Guided Movement and Sound
Attach short reference videos at 480p to steer camera movement and subject motion, and add audio tracks when the clip needs a specific sonic identity. Audio must accompany an image or video reference.
Vertical Storytelling
Ultra-Tall 9:16 and 9:21 Clips
Beyond the usual landscape and portrait ratios, the 9:21 option covers full-height mobile formats, so one API serves cinematic frames and phone-first stories alike.
Notes & FAQ
MiniMax H3 LoRA input and pricing guidance
Notes
01
Available through the APIXO API and the in-browser Playground today.
02
MiniMax H3 LoRA is a separate model from MiniMax H3, with its own endpoint, parameters, and rates: 480p and 768p output instead of 768p and 2K, and every reference image billed at 20 Credits with no free quota.
03
Output bills per second: 40 Credits at 480p and 80 Credits at 768p. Reference images and audio tracks add 20 Credits each; reference videos add 50 Credits per actual second.
04
Duration must be a whole number from 5 to 15 seconds and defaults to 5. Resolution defaults to 480p.
05
Reference videos work at 480p only: up to three clips, each 0–15 seconds, with a combined length of at most 15 seconds.
06
Reference audio (up to three tracks) cannot be used on its own — combine it with reference images or a reference video.
Frequently Asked Questions
Output video costs 40 Credits per second at 480p and 80 Credits per second at 768p. Reference-to-video output uses its own rate of 50 Credits per second at 480p or 100 Credits per second at 768p. Each reference image costs 20 Credits, each reference audio track costs 20 Credits, and reference video costs 50 Credits per second of actual footage. Total price is the output duration times the resolution rate, plus the per-item reference charges.
Reference-to-video accepts up to nine image URLs, up to three video URLs (480p only, each 0–15 seconds, 15 seconds combined), and up to three audio URLs. Audio cannot be submitted alone. Image-to-video takes one or two images instead.
Duration can be any whole number from 5 to 15 seconds. Aspect ratio offers seven options: 21:9, 16:9, 4:3, 1:1, 3:4, 9:16, and 9:21.
Yes. MiniMax H3 LoRA is available through both the APIXO API and the in-browser Playground today.
Explore Other Models
Discover more AI models for your next creative workflow
MiniMax
MiniMax H3
Call MiniMax H3 through the APIXO API for 768p or 2K video with native stereo audio, from text, image, or mixed image/video/audio references. Text-to-video, image-to-video, and reference-to-video from one endpoint. Also known as Hailuo 3.0.
VideoNew
100 Credits/sec
Videohailuo
Hailuo 2.3
Use MiniMax Hailuo 2.3 through APIXO for expressive text-to-video and image-to-video generation with standard 768p and pro 1080p modes.
VideoNew
56 Credits/secAlibaba
HappyHorse
Use Alibaba HappyHorse through APIXO for text, first-frame, multi-reference, and video-edit workflows with audio and 720p or 1080p output.
VideoNew
125 Credits/secByteDance
Seedance 2.5
Explore Seedance 2.5 for longer audiovisual storytelling, timeline direction, multimodal references, multilingual scenes, and selective video revision.
VideoNew
85 Credits/secGoogle
Gemini Omni
Use Gemini Omni on APIXO for multimodal video generation with image, source-video, reusable audio, and character references plus output up to 4K.
VideoNew
100 Credits/secAlibaba
Wan 2.7
Edit and generate videos with Wan 2.7 on APIXO. Supports video-to-video editing, multimodal reference, 1080p resolution, and synchronized audio.
VideoNew
100 Credits/secKuaishou
Kling 3.0 Std
Use Kling 3.0 Standard through APIXO for 3–15-second video generation, optional native audio, multi-shot prompts, and character motion transfer.
VideoNew
84 Credits/secBlack Forest Labs
FLUX 3
FLUX 3 on APIXO supports text-to-video, image-to-video, and video extension workflows with async MP4 output, audio enabled by default, and 5–20 second durations.
VideoNew
170 Credits/sec