A comprehensive AI model platform spanning text, speech, image, and video generation. It offers specialized model families — the M-series large language models, the Hailuo video generation line, Speech text-to-speech, and Music generation — accessible through both an API and web apps. Founded in 2021, this China-based AI company rose to global fame with its hit video generation app “Hailuo AI.” For developers, it provides a console and playground, with a choice between pay-as-you-go API pricing and monthly subscription token plans.
Key Features
- M-series large language models: LLMs focused on agentic workloads, tool calling, and coding. Since MiniMax-M2 in October 2025, the line has iterated through M2.1, M2.5, and M2.7, with MiniMax-M3 arriving in June 2026 with strengthened long-context handling and reasoning
- Hailuo video generation models: The flagship series that generates video from text or images. Following Hailuo 2.3, known for body movement and physical realism, MiniMax H3 (Hailuo 3.0) launched in July 2026, understanding text, images, video, and audio together and generating video with audio in one pass
- Speech text-to-speech: Natural-sounding, multilingual voice synthesis. Options include low-latency variants for voice agents and Speech-2.8 with sound tags for emotional expression
- Music generation: Creates songs from lyrics and style prompts. Each version has expanded its range, including roughly 4-minute song generation and instrumental support
- Both API and web apps: Developers can build with the API, console, and playground, while general users can access generation features from web apps like Hailuo AI right in the browser
- Open-weight releases: Some models are published with open weights on GitHub and Hugging Face, drawing attention for research and self-hosting use cases
Pricing
| Plan | Price | Highlights |
|---|---|---|
| API (pay-as-you-go) | Per-token / per-generation rates by model | Pay only for what you use across the M-series, Hailuo, Speech, and Music |
| Token Plan | $20/month (Plus) / $50/month (Max) / $120/month (Ultra); annual billing discounts available | Plan that bundles a fixed monthly token allowance, usable from MiniMax Code and external coding tools |
| Web apps (Hailuo AI, etc.) | Free tier + paid plans | Use video, speech, and other generation features from the browser |
Specific per-model rates and plan prices are published on the official pricing page. Pricing information is as of August 2026. Please check the official website for the latest pricing.
Pros & Cons
✅ Pros
- Covers text, speech, image, video, and music on a single platform, avoiding fragmented contracts and API integrations
- The Hailuo video series is well regarded for physical realism and human motion, and the latest H3 generates video with audio in a single pass
- The M-series targets agentic and coding workloads, with a fast model update cadence
- Multiple entry points — API, playground, and web apps — make it approachable for both developers and non-engineers
⚠️ Cons
- With so many models and plans, first-time users may find it hard to decide which model and billing scheme to choose
- As a China-based service, enterprise users should review data handling policies and compliance requirements
- Official Japanese documentation and support are limited; most information is in English
- In individual domains, dedicated specialists (video-only or voice-only services) may still warrant comparison
Comparison with Similar Services
| Comparison | MINIMAX | OpenAI | Google Gemini | Kling AI |
|---|---|---|---|---|
| Main scope | Text, video, speech, and music combined | Text, image, video (Sora) | Text, image, video (Veo) | Video generation focused |
| Video generation | Hailuo / H3 (video with audio) | Sora | Veo | Kling |
| Music generation | Music series | Not available | Experimental features | Not available |
| Delivery | API + web apps | API + ChatGPT | API + Gemini app | Web app + API |
| Free tier | Yes | Yes | Yes | Yes |
Who Is It For
- Developers who want to consolidate video, speech, music, and other media generation under a single API contract
- Creators and marketers who want an easy way to try AI video with sound
- Anyone comparing LLM options for agentic and coding workloads beyond OpenAI and Anthropic
- People following open-weight model trends with self-hosting or research use in mind
Summary
MINIMAX is a comprehensive platform where one company covers the major domains of generative AI: the M-series LLMs, Hailuo for video, Speech for voice, and Music for songs. The Hailuo video series in particular has earned global recognition, and 2026’s H3 evolved it into video generation with built-in audio. The breadth of models and billing options can feel complex at first, so a good path is to start with the playground or the web apps’ free tiers, then move to pay-as-you-go API usage or a token plan once your use case is clear.