Last updated: 11/19/2026Last verified: 2026-11-19
Chinese AI platform with 600+ voice models (Speech 2.8 Turbo), multimodal chat, and MiniMax H3 / H3-Image for video and image generation. H3 generates video up to 15 seconds at 2K with native stereo; H3-Image targets 4K stills and up to 16 reference images. Weights have been released; Hugging Face labels the license as other.
What is MiniMax?
MiniMax is a Chinese AI platform focused on audio generation, multimodal chat, and video generation. It offers more than 600 voice models and features Speech 2.8 Turbo for text-to-speech use cases. The platform is positioned for users who need AI voice output alongside chat and media-generation capabilities. MiniMax H3 and MiniMax H3-Image are the current video and image models on that platform: H3 generates video up to 15 seconds at 2K with native stereo sound, and H3-Image targets 4K (4096 x 4096) stills with up to 16 reference images. Both model families are available through the MiniMax API, MiniMax Code and the consumer Hailuo app, and MiniMax has released their weights.
Key Features
- 600+ voice models for text-to-speech and voice generation workflows
- Speech 2.8 Turbo for high-quality AI speech generation
- Multimodal chat capabilities for interacting with different content formats
- Video generation tools for creating AI-assisted visual content
- Freemium access model, based on the provided tool information
- MiniMax H3 video generation: up to 15 seconds per clip at 2K with native stereo audio
- MiniMax H3-Image: 4K (4096 x 4096) text-to-image, reference-based generation and image editing with up to 16 reference images
Best For
Pricing
MiniMax uses a freemium model, and its API is billed pay-as-you-go at rates published on the official pricing overview checked on 2026-11-19: MiniMax-H3 video costs $0.13 per second at 2K and $0.08 per second at 768P, while MiniMax-H3-Image Standard costs $2.30 per million input text tokens, $1.50 per million input image tokens and $3.80 per million output image tokens. The consumer Hailuo app sells separate subscription plans; this listing does not quote its plan prices, so check the official Hailuo pricing page before buying.
Pros & Cons
Pros
- Large library of 600+ voice models provides flexibility for different voice styles and content needs
- Speech 2.8 Turbo highlights a strong focus on text-to-speech quality and performance
- Combines AI audio, multimodal chat, and video generation rather than focusing on a single workflow
- Freemium model may make it easier for new users to test the platform before committing
- Useful for creators and teams working with voice-led or multimedia AI content
Cons
- API usage is pay-as-you-go, so video cost scales with clip length and resolution, and images beyond the free per-task allowance are billed per image
- Users may need to verify language support, regional availability, and API access directly on the website
- The breadth of features may require testing to determine which capabilities are mature enough for a specific workflow
- Security, compliance, privacy and enterprise-support details are not documented in this listing and should be confirmed with MiniMax directly
Alternatives
A well-known AI voice platform for text-to-speech, voice cloning, and audio content creation.
Offers AI text-to-speech and voice generation tools for creators, businesses, and developers.
Provides AI voiceovers for presentations, videos, e-learning, and marketing content.
A multimodal AI chatbot alternative for text, image, and voice-based interactions depending on plan and availability.
An AI video generation and editing platform suited for users focused on visual media creation.
FAQ
Details
Platform
Features
- 600+ voice models
- Speech 2.8 Turbo
- Multimodal chat
- MiniMax H3 video (up to 15s, 2K, native stereo)
- MiniMax H3-Image (4K, up to 16 reference images)
Languages
Known limitations
- API use is pay-as-you-go: H3 video is billed per second of output, so cost scales with clip length and resolution, and images beyond the free per-task allowance are billed per image
- H3 video generation is documented at 4-15 seconds per clip, and H3 / H3-Image are models on the MiniMax platform (API, MiniMax Code and Hailuo) rather than separate standalone products
- Model weights have been released for H3 and H3-Image, but the GitHub repository has no LICENSE file and Hugging Face labels both models as other; the community license agreements exclude the European Union, the United Kingdom, South Korea and the United States
- The consumer Hailuo app sells its own subscription plans, and its live plan prices are not quoted here; check the official Hailuo pricing page before buying








