On March 23, 2026, MiniMax officially launched the Token Plan multimodal subscription plan, completing a comprehensive iterative upgrade of the original Coding Plan. This measure not only breaks the single functional limitation of AI model subscription services, but also opens up a new path for the commercialization exploration of the large model industry with a unified call mode for all modalities.
Full modal integration: a single subscription covers the entire scene of text, images, audio, and video
Core coverage capabilities include:
M2.7 Programming Model: Continuing professional level code generation and optimization capabilities to meet the efficient programming needs of developers
Hailuo video model: supports intelligent video generation and editing, reducing the threshold for visual content creation
Speech model: Achieving high naturalness speech synthesis and interaction, suitable for multi scenario speech applications
Music Model: Quickly generate original music clips, enriching the dimensions of audio creation
Image generation model: precise output of customized visual images to meet the needs of graphic and textual creation
From a single programming tool to a full scenario creation ecosystem, MiniMax has completed its strategic transformation from a vertical tool to a comprehensive productivity platform through this upgrade, allowing developers to complete full chain AI creation without switching multiple subscription services.
Subscription architecture optimization: balancing basic usage with exclusive multimodal quotas
In terms of billing and resource allocation, Token Plan retains the core advantages of the original Coding Plan, while completing targeted upgrades for multimodal requirements.
On the one hand, the platform strictly follows the original 5-hour billing cycle rules to ensure the stable call limit of the M2.7 programming model, so that old users do not need to adapt to the new billing logic. On the other hand, users of Plus and above packages can enjoy independent multimodal calling quotas, which do not occupy the original points of the programming model and enable bidirectional independent use of programming and multimodal capabilities.
In response to the high concurrency and large-scale calling needs of professional developers, MiniMax has launched a dedicated resource package that is compatible with Speech2.8 flagship voice model and Hailuo2.3/2.3-Fast video model. Compared with the traditional pay as you go model, it can help users reduce usage costs by about 20% and further enhance the cost-effectiveness of enterprise level development.
Traffic regulation mechanism: balancing peak experience and ultra-high concurrency demand
With the continuous increase in user calls after the M2.7 model was launched, MiniMax introduced an industry wide dynamic control mechanism to ensure the overall stability of the platform operation.
The platform implements a reasonable flow restriction strategy during peak working hours to avoid service delays caused by traffic congestion; At the same time, we provide solutions for tasks with ultra-high concurrency requirements and recommend users to switch to the pay as you go API mode. This mode is not limited by flow restrictions and can meet the rigid requirements of large-scale batch calls, catering to the different usage scenarios of ordinary users and professional developers.
Industry value: paradigm upgrade from technological breakthroughs to integrated services
The implementation of Token Plan marks the development focus of top model manufacturers, shifting from a single technical parameter competition to a deep commercial cultivation of full scenario service integration.
By integrating the multimodal capabilities of text, graphics, audio, and video into a unified subscription framework, MiniMax significantly reduces the technical and cost barriers for developers to build complex AI agents, allowing small and medium-sized teams to quickly build full modal intelligent applications. This mode also provides a reference commercial paradigm for the big model industry, and promotes AI technology to be transferred from the experimental call in the laboratory stage to a practical productivity tool for office, creation, development and other scenarios.
In the future, with the popularization of multimodal subscription models, the threshold for using AI services will continue to decrease, and more industries will leverage unified AI capabilities to achieve efficiency upgrades in digital creation and operation.
