# MiniMax

## Kurzbeschreibung
**"co-create intelligence with everyone"**


MiniMax is a multimodal foundation model provider with models for text, coding, agents, speech, video, music, image, and multimodal applications. Its products include MiniMax Agent, Hailuo AI, MiniMax Audio, Talkie, and an Open Platform for developers and enterprises.

## Claim
LLM “co-create intelligence with everyone”

## Geeignet für
- API Integration
- Audio / Voice
- Automation / Workflows
- Images
- Design / Graphics
- Customer Service & Chatbots
- Programming / Software Development
- Social Media
- Texts / Content
- Video

## Kernfunktionen
- Image generation
- Chatbot
- AI agents
- LLM API
- Multimodal AI
- Music Generation
- open-source model
- Programming
- Voice output
- Language model
- Voice cloning
- Text generation
- Video generation

## Preismodell
- **subscription:** **Token Plan – Starter** Subscription access for developers with access to MiniMax models via Token Plan API key; text models with a 5-hour rolling window, plan-dependent limited multimodal quotas.


**Token Plan – Plus** Expanded Token Plan with a higher M2.7 quota and additional daily multimodal quotas for Speech, Image, and other models.


**Token Plan – Max** Higher standard Token Plan with a larger request quota and expanded daily quotas for multimodal models.


**Plus-Highspeed / Max-Highspeed / Ultra-Highspeed** High-speed Token Plans with access to faster M2.7/M2.5 Highspeed models and expanded quotas for coding-related workflows.
- **other:** **Pay-as-you-go API** Standard Open Platform API key for usage-based billing according to actual consumption; supports text, video, speech, image, and additional modalities.


**Audio Subscription / Video Packages** Separate packages for speech and video generation with product-specific quotas and billing logic.


**Local / Private Deployment** Open M2 model weights can be run locally or privately via Hugging Face and frameworks such as SGLang, vLLM, Transformers, ModelScope, or NVIDIA NIM.

## DSGVO und Datenschutz
**Gesamteinschätzung:** unklar / nicht verfügbar

**On-prem / local hosting: supported**

Supported for MiniMax-M2.7: There is a documented "Local Deployment Guide" for Linux GPU servers and Mac Studio, including fully local operation on Apple Silicon hardware.

**Private Cloud / Data Center: Indirect / Not Available**

There are references to enterprise/API integrations in the pricing overview, but no specifically described dedicated private cloud, VPC, or EU data center options on the website.

**EU SaaS / Managed: unclear**

SaaS/API usage is clearly documented, but EU/EEA data residency, EU data centers, or a European hosting option are not specified on the website.

**Hybrid: Partially**

Partially inferable, as on-premises deployment is documented for MiniMax-M2.7 and self-hosted MCP services, while other features are offered as API/SaaS. However, an explicit hybrid architecture or an officially described combination of on-premises and hosted processing is not specifically guaranteed.

**DPA: unclear**

Although the Terms of Service mention “data privacy and security agreements” as part of the Service Rules, no specific DPA or process for entering into one is provided on the website.

**No training: partial**

Partially, and only for a specific area: The API documentation for the synchronous text-to-speech interface states that the interface is stateless and “the model does not store any user data.” A general, contractual prohibition on using prompts, uploads, chat histories, or outputs to train general models is not specified on the website.

**Open Source / Transparency Path: Partially**

There is a transparency/sovereignty path: The website documents local deployment of MiniMax-M2.7 and refers to MiniMax M2 on the website as an open-source model context. In addition, official MCP servers in Python and JavaScript, as well as self-hosting of these MCP services, are described. However, the website does not fully specify which components are open source overall.

**Data Processing**

For standard use, processing via the MiniMax API is documented. Files can be uploaded to the platform, and task and file management functions are described for asynchronous speech and video processing. For TTS Async, it is described that generated audio files are made available via a time-limited URL and may be lost after the expiration date; for Voice Cloning, temporary voices are described that are automatically deleted after 7 days if they are not used. However, for the European region, there is no information on in which country or region this processing takes place or whether EU/EEA data residency is available.

**Conclusion**

The best way, as documented on the website, to achieve greater European data protection sovereignty is through the **local deployment of MiniMax-M2.7** or self-hosted MCP components. For regular SaaS/API use within the EU/EEA, however, key GDPR compliance evidence is missing. Therefore, the overall assessment for the European region is *unclear* and not “yes.”

**Sources**

- [https://platform.minimax.io/docs/api-reference/api-overview](https://platform.minimax.io/docs/api-reference/api-overview)
- [https://platform.minimax.io/protocol/privacy-policy](https://platform.minimax.io/protocol/privacy-policy)
- [https://platform.minimax.io/protocol/terms-of-service](https://platform.minimax.io/protocol/terms-of-service)
- [https://platform.minimax.io/docs/guides/local-deploy](https://platform.minimax.io/docs/guides/local-deploy)
- [https://platform.minimax.io/docs/guides/mcp-guide](https://platform.minimax.io/docs/guides/mcp-guide)
- [https://platform.minimax.io/docs/pricing/overview](https://platform.minimax.io/docs/pricing/overview)
- [https://platform.minimax.io/docs/llms.txt](https://platform.minimax.io/docs/llms.txt)

For the European market, it is *unclear* whether the website is GDPR-compliant. On the positive side, a privacy policy with EEA/UK/Switzerland-specific information has been published, and there is a documented local deployment path for MiniMax-M2.7. However, for SaaS/API use, the website lacks essential evidence such as EU/EEA data residency, server locations, a readily accessible Data Processing Agreement (DPA), a list of subprocessors, a clear contractual exclusion of model training using customer data, and robust compliance certifications.

**Positive**

The website features a published privacy policy with a reference to “jurisdiction-specific provisions” for users in the EEA, UK, and Switzerland. In addition, a local deployment path for MiniMax-M2.7 is documented; there, “complete data privacy” is explicitly described for Mac Studio when operating entirely locally. Furthermore, developers can host MCP services themselves.

**Negative**

The website does not specify an EU/EEA server location, EU data residency, a specific, accessible text of the AVV/DPA, a list of subprocessors, an explicit opt-out from AI training for general models, or ISO 27001, SOC 2, or similar certifications. The Terms of Service name Nanonoble Pte Ltd. as the provider and refer to Singaporean law; this does not replace GDPR compliance documentation for EU/EEA use.

**Server location**

Not specified on the website. API endpoints are documented under 'api.minimax.io', but there is no information on the location of servers or data centers, nor is there any mention of EU/EEA data residency.

## Hosting und Daten
- **On-Prem / lokales Hosting:** abgedeckt
- **Private Cloud / Rechenzentrum:** unknown
- **EU SaaS / Managed:** unknown
- **Hybrid:** teilweise / indirekt
- **AVV / DPA:** unknown
- **Kein Training auf Kundendaten:** teilweise / indirekt
- **Open-Source / Transparenz-Pfad:** teilweise / indirekt

## Standort
**Land:** China

**Taxonomie:** China

Xuhui District in Shanghai.

## Vorteile
- Very broad multimodal portfolio.
- Strong text/coding/agent models.
- Audio, video, and music models from a single platform.
- Open-weight models and local deployment paths for multiple models.

## Nachteile
- GDPR documentation is only publicly assessable to a limited extent.
- No publicly verified EU-only operation.
- No publicly verified DPA/AVV in the sources reviewed.
- Multimodal data such as voice, image, and video increase data protection risks.

## Quellen
- Offizielle Website: https://platform.minimax.io/docs/api-reference/api-overview

## Letzter Datenstand
2026-05-06

## Originalseite
https://kifox.ai/en/ki-tools/minimax-en/