LLM Provider

LLM Provider defines the large model connection method used by ASP. AI investigation, knowledge extraction, and Runtime select models from enabled Providers by tags and priority.

View

LLM Providers list displays Name, Base URL, Model, Tags, Enabled, and Priority.

The list supports quick filtering by Enabled and Tags, and also supports advanced filtering by Name, Base URL, Model, Enabled, and Priority to locate configurations.

Fields

Field	Description
Name	Configuration name, unique.
Base URL	OpenAI Chat Completions compatible model service address.
Model	Model name.
API Key	Access key, can be filled according to model service requirements.
Proxy	Optional proxy.
Tags	Model capability tags.
Enabled	Whether enabled.
Priority	Priority, smaller number means higher priority.

Add and Edit

Admins can add, edit, enable, disable, or delete LLM Providers. When editing, saved configuration is read; API Key is hidden by default, only loaded through reveal when editing.

Proxy supports addresses starting with http://、https://、socks4://、socks5://。

Tag Selection

Common tags:

• fast
• powerful
• tool_calling
• structured_output

At least one tag must be filled. The frontend provides common tags, but you can also add custom tags based on actual scenarios.

Different tasks select appropriate models based on tags. For example, structured output tasks prefer Providers tagged with structured_output.

Test Connection

When adding or editing configuration, you can directly Test; saved Providers can also be Test from the list.

The test will call the Provider's Chat Completions compatible interface and send a simple request to confirm model service is available.

Runtime Selection

Only Enabled Providers enter runtime configuration. Runtime sorts by Priority, Name, Created Time:

• When no tag is specified, use the first Provider after sorting.
• When one tag is specified, select the first Provider containing that tag.
• When multiple tags are specified, select the first Provider containing all tags.

If there are no enabled Providers, or no Providers matching specified tags, related AI tasks will fail.

Security and Audit

Creating, updating, deleting, testing, and revealing API Key are all written to Audit Log. API Key fields in audit records only record whether a value changed or was revealed, not plaintext secrets.

Usage Recommendations

• At least configure one enabled structured_output Provider for investigation reports and knowledge extraction.
• Use Priority to control default model selection order, smaller number means higher priority.
• Use Tags to distinguish fast models, strong reasoning models, tool calling models, and structured output models.
• After saving, first execute Test to confirm Base URL, Model, API Key, and Proxy configuration are correct.

Next Steps

• Runtime — Configure Prompt Language and runtime parameters.
• Case — See how LLM Provider supports case investigation reports.
• Knowledge — See how structured output is used for knowledge extraction.