Model List is the entry point in the DataBuddy data application module for centrally aggregating large language model (LLM) endpoints. Here, you can view the out-of-the-box foundation models hosted by the platform, which can be called by modules such as Agent, MCP, Notebook, SQL Editor, and Buddy in a unified manner.
Module Positioning
The Model List addresses the fundamental question of "where LLMs are and how to call them":
For Agent developers: you only need to write a single line model="wedata-gpt-5" in your Agent code, without maintaining API keys, Endpoint URLs, timeout, and retry logic in your project.
For data analysts: directly call functions such as ai_query() in Notebook or SQL Editor, treating the LLM as a regular SQL UDF.
For administrators: all LLM calls are routed through platform-managed Endpoints, facilitating unified monitoring of call volume, auditing of call records, and control of permissions and quotas.
Tip
Model List vs. Data Science Model Management
The Model List (7.3) is intended for large language model (LLM) endpoints called by Agent / applications, and is an asset of the data application module.
Model management in data science is intended for machine learning models (ML Models) registered with MLflow, which are trained by data scientists and registered to Catalog.
The two are independent lists in the product, and this document describes only the former.
Available Models
The platform pre-provisions and hosts mainstream LLMs, allowing users to use them out of the box without configuring API keys or credentials.
The currently available platform-hosted base models include (continuously updated):
wedata-gpt-5 series
wedata-deepseek-v3 series
Other platform-hosted models (see the Model List page for details)
To learn about the capability differences, billing rules, and applicable scenarios of each model, click the corresponding model on the Model List page to view details.
Key Concepts
To understand the Model List, you need to be familiar with the following concepts:
Model: a specific LLM instance, such as wedata-gpt-5 or wedata-deepseek-v3. Each model corresponds to a unique model ID, through which the model is called in code.
Model ID: the unique identifier for calling a model within the DataBuddy platform, following the wedata-<provider>-<model-name>-<version> naming convention, for example, wedata-gpt-5-2. The model ID is the string you need to write in your code.
Endpoint: the unified call address generated by DataBuddy for each model. You do not need to directly care about which cloud or region the backend model is actually deployed in, as the platform handles routing and failure retries.
Platform-hosted: Models are centrally managed by the DataBuddy platform, including deployment of model instances, API keys, quotas, allowlists, and monitoring.
Allowlist: Platform-hosted base models can be used only after the allowlist is enabled. After it is enabled, usage is recorded in audit logs.
Features
Capability | Description |
Browse Model List | View all available models on the Model List page under the Data Application module, with support for searching by name. |
Copy Call ID | Each model card supports one-click copying of the model ID, which can be pasted into Agent code / Notebook for invocation. |
Access Documentation and Billing Information | Each model provides entry points to the vendor's official documentation, copyright notice, and billing information, helping you evaluate whether the model meets your business requirements. |
Call Audit and Usage Statistics | All calls to platform-hosted models are routed through a unified Endpoint, call logs are traceable, and usage is measurable (for billing policies, see 7.3.4 Model Invocation and Billing, to be supplemented). |
Counterpart Products
DataBuddy | Databricks | Alibaba Cloud DataWorks / Bailian |
Model List / Platform-Managed Foundation Models | Foundation Model APIs(Pay-per-token) | Built-in models of Bailian |
Model ID (such as wedata-gpt-5) | Endpoint name (for example, databricks-gpt-5) | Unique model code |
Use Cases
Models in the model list can be called in the following scenarios:
|
Agent | Agent frameworks (OpenAI / LangChain / LangGraph) obtain model instances through DataBuddy SDK. | |
Notebook | Call directly through OpenAI SDK / LangChain / DataBuddy SDK. The URL is the model endpoint address. | |
SQL Editor | Reference the model ID through built-in SQL functions such as ai_query(). | - |
Buddy | Buddy automatically selects a model in the backend, and you can also switch it at the bottom of the dialog box. | |
Sub-Feature Navigation
Sub-feature | Description | Documentation |
View available models. | Browse model list, view model details, copy model ID, and access vendor documentation. | View available models. |
Model Invocation and Billing | How to call models in code, and billing rules | 7.3.4 Model Invocation and Billing (To Be Supplemented) |
Recommended Learning Path
If you are using the model list for the first time, read the following sections in order:
1. Learn Module Positioning: this section.
2. Familiarize Yourself with Available Models: Find the model ID that suits your business in View Available Models.
3. First Call: Refer to 7.3.4 Model Invocation and Billing (to be added) and follow the code example to complete a model call.
References