Quick BI supports flexible configuration of large models. You can select and switch between different large models within intelligent features such as Q-Copilot, Q-Insight, Q-Report, and Q-Dash for various intelligent interaction and data analysis scenarios. This topic describes how to configure large models.
Entry point
Follow the steps in the figure to open the Large model configuration page.
Built-in large models
Quick BI currently supports the following built-in large models: Deepseek-V4-PRO, Qwen3.6-plus, Qwen3-max, and DeepSeek-R1-0528.
The built-in large models are provided by Quick BI. To ensure data security, Quick BI does not log or store any data during their use.
Token quota
The intelligent modules primarily rely on the built-in large model, and using this model consumes Tokens. We currently provide a certain number of free Tokens to paid users of the Advanced and Professional Editions. For more information, see Free Token and Consumption Rules. When the token quota is exhausted, you can no longer use the built-in large model. You can then choose to purchase a token top-up package or add a custom large model.
Built-in large models are optimized for the best performance across all features. You do not need to purchase a separate model; you only need to purchase a token top-up package.
Custom large models
Add a custom large model
To add a purchased third-party large model, follow these steps.
Click Add.
Select a service provider.
You can choose either Model Studio (Bailian) or an AI gateway. Configure the parameters based on your selected provider.
Model Studio (Bailian)
Parameter
Description
Service provider
Select Model Studio (Bailian).
Model name
Select an available model from Model Studio (Bailian) or enter a custom model name.
Context length limit
The maximum context length supported by the model.
Display name
A custom display name for the large model. The name can be up to 50 characters long.
API key
For instructions on how to obtain an API key, see Get API Key.
AI gateway
Parameter
Description
Service provider
Select AI gateway.
AI gateway domain
Enter the domain name of your AI gateway, for example,
https://gateway.example.com.Model name
Enter a custom model name.
Context length limit
The maximum context length supported by the model.
Display name
A custom display name for the large model. The name can be up to 50 characters long.
API key
For instructions on how to obtain an API key, see AI Gateway.
Click Test connection to test the connection to the large model.
After the test is successful, click Confirm to save the configuration.
Once added, you can set the model as the default or use it in features like Q-Copilot and Q-Insight.
Manage custom large models
You can manage a custom large model by enabling (①), disabling (②), editing (③), or deleting (④) it.
① If the model is disabled, click the 
icon to enable it.
② If the model is enabled, click the 
icon to disable it.
③ Click the 
icon to edit the model. You can only modify the display name.
④ Click the 
icon to delete the model.
Default model
You can set one of the large models as the default. This improves efficiency by avoiding the need to select a model for each use.
Behavior:
When the default model applies
After a default model is set, it is automatically selected in scenarios that allow model switching, such as Q-Copilot, Q-Insight, and Insight Analysis on the Q-Report editing page.
For example, if you set DeepSeek-R1-0528 as the default large model,
Q-Insight selects it by default. You can switch to another large model as needed. When the default model does not apply
In scenarios that do not allow model switching, the default model setting does not apply. For example, when you use the Generate smart report with one click feature, Quick BI automatically uses multiple models to ensure the best report quality.
Q-Insight selects it by default. You can switch to another large model as needed. 