This document provides answers to frequently asked questions (FAQs) about Alibaba Cloud Model Studio.
Billing
What are the unit prices for the models in Alibaba Cloud Model Studio?
See Models and Billable items.
Are subscription services available?
Some models support subscription, see Savings plans.
Are pay-as-you-go bills settled monthly?
Bills are generated hourly and settled monthly.
How to query my charges and view details?
You can go to Expenses and Costs to view the details.
How to request an invoice for my expenses?
On the Overview of Monthly Bill page, click Download Invoice in the Actions column next to the destination account.
API/SDK
How to view error code information?
An API call to Model Studio returns a status code that indicates the result of the call. For more information about error codes and their solutions, see Error messages.
How to install the SDK?
Model Studio currently supports SDKs for Java and Python. For more information, see Install the SDK.
When calling a function with the assistant API, does it support calling two local functions in sequence?
a. Calling two functions separately is not currently supported.
b. You can manually create two separate assistant APIs and handle the response from each one individually.
Does the assistant API have memory-related capabilities?
The memory configuration feature is not currently supported.
Product
How to activate the Alibaba Cloud Model Studio service?
Log on to your Alibaba Cloud account and go to Alibaba Cloud Model Studio (Singapore) or Alibaba Cloud Model Studio (Beijing). On the Homepage or Models page, you can activate the model service. For more information, see Product activation.
How to deactivate the Alibaba Cloud Model Studio service after activation?
The Model Studio service cannot be deactivated. If you use the API to call models or applications, you can delete your API key on the API-KEY (Singapore) or API Key (Beijing) page in the console to prevent subsequent calls.
How to experience the model service?
You can experience the service in the Playground (Singapore) or Playground (Beijing).
What is the difference between Alibaba Cloud Model Studio and Qwen?
Alibaba Cloud Model Studio is a model service platform that provides various models, including the Qwen series.
How can I implement data isolation to ensure that data from different users is not associated?
You can use an Alibaba Cloud account to grant different workspace permissions to different RAM users. This practice ensures that data in different workspaces is isolated. For more information, see Workspace Management.
Does Model Studio save the data generated during model calls?
Alibaba Cloud strictly protects your data privacy and will never use your data for model training. In addition, all data that you transmit when you build applications is encrypted to ensure data security.
When you call a model through Model Studio, how your data is processed and retained depends on how you call the model:
When you call a model through a model API, Model Studio does not save conversation data. During the call, only desensitized logs of the call status are recorded. After the call ends, the plaintext prompts and responses are destroyed, and no data is saved.
When you call a model through the assistant API, Model Studio saves conversation data to enhance the user experience. Currently, there is no expiration date for this data.
When you call a model through a Model Studio application, the conversation history is cached by default for 60 minutes. The number of cached conversation turns is specified by the 'Number of context turns' parameter. If a new message is sent within 60 minutes, the cache expiration time is extended by another 60 minutes. If no new messages are sent within 60 minutes, the conversation history expires and the data is destroyed.
Models
How are the parameters of a model stored?
You can download open source models from the ModelScope community. Their structure is usually defined in a JSON file. You typically need to use open source Python libraries to parse these files. These files contain vector information, and parsing them can help you understand the storage process.
How many languages do the Qwen series models support?
14 languages: Chinese, English, Arabic, Spanish, French, Portuguese, German, Italian, Russian, Japanese, Korean, Vietnamese, Thai, and Indonesian.
Can the current models connect to structured data, such as MySQL and Hive?
This is not currently supported. However, this feature is under development, and priority is being given to connectivity with ApsaraDB RDS services.
Is the text generation speed of models such as Qwen3 and Qwen-Max fixed for all users, or is there a way to adjust the speed?
The generation speed is not fixed. It is affected by factors such as the overall load of the current service and your request concurrency.
After model throttling is triggered, how long should I generally wait before trying again?
The waiting time depends on your specific rate limit value, such as requests per second (RPS) or requests per minute (RPM). For example, if your limit is 120 RPM (2 requests per second), and you submit 2 requests consecutively within 0.2 seconds, the 3rd request will be throttled. You will need to wait about 0.8 seconds before you can successfully submit another request.