Alipay Token Pay integrates MiniMax M3, the latest domestic AI model

MiniMax's latest large model, M3, is now fully integrated into Alipay's TokenPay one-stop full-stack solution, allowing enterprises to enjoy bulk purchasing and one-click distribution for managing Token seats. Users can also complete in-app payments through Alipay's AI Pay by simply speaking a command within the Mix Claw service powered by MiniMax Agent.

MiniMax is one of the first global large model developers to embed Token Pay payment capabilities directly into its core subscription system, marking the first large-scale deployment of Alipay's domestic MaaS (Model as a Service) payment solution.

MiniMax is also set to fully integrate with Alipay's full-stack AI payment products in the near future, upgrading the service experience for both developers and individual users, accelerating commercialization, and enabling developers to use natural language to call payment capabilities, making it easier for users to instruct Agents to perform tasks more reliably and obediently.

Enterprise-Grade Token Seat Management

Bulk Purchasing with One-Click Distribution

As large language models become enterprise productivity tools, more companies are buying Token services in bulk. However, in practice, they face several management challenges: determining the right number of seats, selecting the appropriate package for each seat, distributing them efficiently internally, and handling reimbursement.

To address these pain points, Token Pay has enhanced its enterprise seat management features, including one-click subscription for enterprise groups, the ability to scale seats up or down, and support for complex scenarios like group management and token allocation. By integrating with MiniMax, this effectively creates a complete payment SaaS system, significantly improving the purchasing experience for enterprise customers.

Deep Integration with Large Language Models

M3 Model Demonstrates Robust Security in Payment Scenarios

This collaboration goes beyond product-level integration, achieving a deep optimization of large model intent understanding and financial-grade payment execution at the foundational level.

On June 1, MiniMax officially launched its new general-purpose model, MiniMax M3, featuring a proprietary sparse attention architecture called MiniMax Sparse Attention (MSA). M3 is the first large model in China to combine three core capabilities: state-of-the-art coding ability, a 1-million-token ultra-long context, and native multimodal support. It is also the only open-source model globally to offer this complete set of features.

During adaptation testing, Alipay observed that M3 excelled in intent recognition accuracy and risk boundary awareness, showing clear advantages in understanding the payment task chain, defining payment security boundaries, and managing engineering complexity. This positions M3 as key infrastructure for achieving a commercial closed loop in AI-native applications.

Low-Code Access

Commercialization Barriers Near Zero

"We've componentized all payment solutions so they are ready to use out of the box," said Zhu Lin, General Manager of Alipay's AI Payment. "In large model commercialization scenarios, developers can apply Alipay's capabilities as needed, allowing them to focus on improving model performance and optimizing underlying computing power."

With low-code integration, various billing models—including subscription, pay-as-you-go, and top-up packages—can be flexibly combined like building blocks. This AI payment solution not only significantly cuts R&D costs for large model companies and boosts payment conversion rates but also serves as a powerful engine for business growth.

A New Payment Paradigm

Becoming the 'Standard Foundation' for Large Model Commercialization

As the number of agents grows exponentially and tokens become the hard currency of the AI economy, the payment paradigm shift is just beginning.

Token Pay's three-layer architecture tackles the core pain points of AI commercialization. The settlement layer ensures agents don't run out of power during task execution. The management layer enables large model makers to purchase and manage tokens for enterprises in a single operation, creating a major efficiency lever. The open layer lowers the access threshold to an extremely low level, allowing various billing models to be flexibly combined like building blocks and accelerating the commercialization process.

Alipay has revealed plans to further explore innovative models like pay-per-performance and dynamic pricing, as well as support unified settlement across agent frameworks. Earlier, Alipay announced a full-stack AI-native payment product system that supports 300 million AI intelligent entities and 95% of general intelligent entity frameworks, making it the world's first large-scale commercial AI-native payment infrastructure.