tencent cloud

TokenHub Migration Guide

Download
Mode fokus
Ukuran font
Terakhir diperbarui: 2026-05-13 14:42:32
To enhance the user experience with large models, the large model sales portal of the original Tencent HY and LLM Knowledge Engine Basic API— DeepSeek API (hereinafter referred to as: the original platform) will be migrated to TokenHub. No new model capabilities will be added to the original platform, and new purchases of model services will be discontinued. For details, please refer to the official platform decommissioning announcement. We recommend that you migrate your purchased model services to TokenHub for use.

Migration Benefits

TokenHub supports a wider selection of more recent and popular industry models, such as DeepSeek V4 Pro/Flash, GLM 5.1, Kimi-K2.6, and MiniMax-M2.7, compared to the original platform. For details, see the Model list.

Scope

1. Users with Pay-as-you-go Enabled on the Original Platform

Users who have enabled Pay-as-you-go billing for Hunyuan and DeepSeek can migrate directly to TokenHub. For the operational steps, see Migration Method.

2. Users with Prepaid Enabled on the Original Platform

TokenHub will no longer support the purchase of prepaid Token resource packages. You can choose to switch to TokenHub after your existing resource package is depleted, or contact the platform for assistance to switch to TokenHub after a refund proportional to the unused resources is received.
Note:
Refund calculation method: Refund Amount = Resource Package Purchase Amount * (1 - Resource Package Usage Ratio). Refunds are typically returned via the original payment method within several days to one week. For detailed information, refer to the refund application review notification.
For users who have purchased with a discount, the platform will assist in transferring the corresponding discount to TokenHub.
The Hunyuan-translation model TokenHub will no longer be supported. When you switch to TokenHub, we recommend that you use a newer model instead.
The online search capability of the original platform will be decommissioned along with the platform. TokenHub will launch a new online search capability, which is a different product from the original platform. Its API protocol, usage method, and pricing have all changed. Related information will be published upon launch.

Migration Method

Step 1: Log in to the TokenHub console, navigate to the Online Inference page, select the model you need, and Enable Pay-as-you-go. After enabling, you will be billed postpaid based on Token usage. For model postpaid pricing, see Model Pricing.

Step 2: Create a new API Key on the TokenHub API Key page. For detailed operations, see API Key.

Step 3: Refer to the invocation sample on the Model Details page (using DeepSeek V4 Flash as an example) to replace the URL and API Key for calling the DeepSeek API.
If you access via the OpenAI SDK, replace the base_url with: https://tokenhub-intl.tencentcloudmaas.com/v1.
If you access directly via URL, the URL is: https://tokenhub-intl.tencentcloudmaas.com/v1/chat/completions.

For more API invocation examples, see the API Usage Guide.

Contact Us

If you encounter any issues or need assistance during the migration process, you can contact us by consulting Online Customer Service or submitting a ticket.



Bantuan dan Dukungan

Apakah halaman ini membantu?

masukan