AI Capability Extension Pack: Subscription Service Description

The AI Capability Extension Pack is an in-app subscription service designed for end users of AI-powered devices. This topic describes the positioning, billing principle, partner benefits, and cooperation models of the service, helping device brands (PID owners) determine whether to incorporate it into the business model of their AI hardware products.

Service positioning

The AI Capability Extension Pack provides a business model in which end users pay for AI resource consumption. After consuming the daily free AI usage quota provided with the device, users can subscribe to the AI Capability Extension Pack to obtain a higher daily AI usage quota.

For device brands, this service divides the AI hardware experience into a free version and a paid subscription version. This helps you better manage the ongoing resource consumption costs incurred during extended use of AI hardware. It also simplifies the consumer understanding and payment thresholds caused by the differentiated cost structures of different AI services.

Billing principle

The AI Capability Extension Pack provides the following features to address complex, differentiated cost structures and billing challenges arising from diverse AI services and user habits.

Unified measurement across billing items

AI service providers typically use different billing metrics, such as conversation duration, character count, token consumption, generated image count, or API call volume. With this service, different types of resource consumption are converted into deductions based on the actual cost ratio of each interaction. This value can be customized according to your product branding, such as Energy Points, Inspiration Points, Stamina Points, Power Points, and so on.

Compatible with all AI hardware capabilities

The subscription service provides a unified resource quota rather than fixed usage limits for each capability. Whether the primary resource consumption of a device comes from voice interaction, recognition, web search, generative AI, or a combination of multiple capabilities, all usage can be deducted from the same quota. This also applies if the device continues to update more capabilities after users subscribe. Users do not need to pay repeatedly for different capabilities. This improves the payment experience and continuous subscription retention.

Flexible quota adjustment

The service can be extended into different versions based on different product types and AI usage scenarios. It can also be continuously optimized based on accumulated cost and usage data, achieving a balance between user experience and cost control.

Continuous recovery and real-time dynamic calculation

For products that adopt the Subscription Service Model, both free and paid usage quotas recover continuously over time. When the quota is fully consumed, users receive a friendly reminder indicating that the device needs to take a break to control the usage frequency in a softer way. The system also calculates and deducts the actual resource consumption generated by each conversation in real time to ensure accurate usage control.

Current service offering

Take the Tuya all-in-one app as an example. To ensure a unified user experience, all AI hardware maintains a unified quota grant when purchasing the same AI Capability Extension Pack. Different AI models and services affect only the deduction amount for each interaction and do not affect the overall subscription cost.

The following table describes the benefits included in the AI Capability Extension Pack – Basic Plan. More subscription plans with different quota levels will be introduced in the future.

Based on the AI-related service cost changes in different data centers, the benefit content of the subscription service may be further optimized over time.

If you use a self-developed app or an OEM app, you can also offer the same subscription service through a distribution or redemption model. For details, see Cooperation models. Custom quota packages are also available upon request.

Deduction examples

Each AI interaction that a user initiates on the device is called a round of conversation. The system converts this into a deducted value based on the actual consumed tokens, whether automatic speech recognition (ASR) or text-to-speech (TTS) is included, and the number of calls to other paid services.

Because pricing varies across AI model providers, paid capabilities, and future vendor pricing adjustments, the deduction amount for each conversation round may differ significantly. For example:

In a round of conversation, the LLM input is 10,000 tokens, the output is 20,000 tokens, the ASR duration is about 10 seconds, and the TTS is about 50 characters. Web search and voiceprint recognition are triggered once at the same time. The system dynamically calculates the combined service cost based on current model and service pricing and converts it into a deduction value according to a fixed point conversion ratio.
If the calculated cost of the interaction is ¥0.04 and the current conversion ratio is 1 point = ¥0.004, the system deducts 10 Inspiration Points.

Partner benefits

For all products (PIDs) enrolled in the Subscription Service Model, devices automatically use the free version usage quota limit and support purchasing the AI Capability Extension Pack – Basic Plan. In addition, all AI resource consumption generated by devices under these products is waived in the platform’s AI agent billing statistics, and you do not need to bear the cost.

Products that are not enrolled in the Subscription Service Model have no usage restriction. AI resource consumption generated by these devices is calculated to the account associated with the product after applicable fee waivers are applied, and you will need to pay based on actual usage volume.

Cooperation models

Using Tuya’s all-in-one app

To ensure a consistent user experience across the Tuya all-in-one app, Tuya provides a unified subscription service for all AI hardware products. You only need to enroll the product (PID) in the subscription model and ensure that the device panel supports navigation to the subscription purchase page. Products using Tuya’s standard AI panel include this capability by default.

If the basic plan does not match your product’s actual usage requirements, Tuya will continue to introduce more suitable plans whenever possible.

Using self-developed or OEM app

You must first choose either the distribution model or the redemption model to integrate the subscription service into your app.

• Distribution model: Tuya directly sells subscription services within your OEM app.
• Redemption model: You purchase subscription packages in advance and determine your own pricing and sales strategy.

Under either model, you are responsible for implementing the subscription purchase workflow within your app.

• For distribution partnership agreements, visit: AI Capability Subscription (Revenue Sharing Model)

• For redemption purchases, visit: China Data Center | Other Data Centers

For subscription pricing, revenue-sharing arrangements, and other commercial terms, contact your Tuya business representative.

Other related services

In addition to recurring subscription plans that provide continuously recovering daily usage quota, the platform also offers one-time consumable items that instantly restore available quota. This allows consumers to recover usage at a lower cost in emergency scenarios.

Other AI-related services that are independent of the number of device conversations, for example, creating and saving cloned timbres, are available through separate purchase channels. These are not included in the subscription service.

For cooperation needs, contact us for further discussion.