Quick Signup

• SiliconFlow: One-click Signup

Purpose

Configure the AI service, API type, model parameters, and the active configuration.

Steps

1. Open the settings page.
2. Enter AI Configuration.
3. Create a new configuration or edit an existing one.
4. Choose the API type.
5. Fill Base URL, API key, and Model name.
6. Adjust Max tokens, Temperature, and Memory rounds if needed.
7. Tap Test Connection.
8. Set the working configuration as active.

Field Guide

Field	Description
config name	configuration name used to distinguish providers or models
Base URL	API base endpoint
API type	request protocol type
API key	access credential
Model name	model ID
Max tokens	single-response limit
Temperature	output randomness
Memory rounds	context window rounds

How To Fill The URL

Base URL should be the API base endpoint. Do not use the provider home page or a documentation page.

OpenAI-Compatible API

Use the OpenAI-compatible base endpoint provided by the service. This is usually a base path or a compatibility endpoint ending with /v1.

Reference examples:

• https://api.openai.com/v1
• https://api.deepseek.com/v1
• https://dashscope.aliyuncs.com/compatible-mode/v1

Do not paste full request paths such as these unless the provider explicitly requires them:

• /chat/completions
• /responses

Anthropic API

Use the Anthropic base endpoint or the Anthropic-compatible endpoint provided by the service.

Reference examples:

• https://api.anthropic.com
• the Anthropic-compatible endpoint from your provider

Do not paste a documentation URL, console URL, or a full business request path into Base URL.

Example Values

API type	Base URL example	Model name example
OpenAI-compatible	https://api.openai.com/v1	gpt-4.1-mini
OpenAI-compatible	https://api.deepseek.com/v1	deepseek-chat
OpenAI-compatible	https://dashscope.aliyuncs.com/compatible-mode/v1	qwen-plus
Anthropic	https://api.anthropic.com	claude-3-5-sonnet-latest

Check These First When Connection Fails

1. whether the API type matches the service
2. whether Base URL is a real API endpoint instead of a website URL
3. whether Base URL incorrectly includes /chat/completions, /messages, or another full request path
4. whether Model name matches a model ID actually supported by the service
5. whether API key contains spaces, line breaks, or a wrong prefix

Parameter Notes

• use a lower Temperature for reverse engineering and code generation
• increase Max tokens when output length is insufficient
• increase Memory rounds when context is too short

Contributors: 挽风

Previous pageProject Management

Next pageUI and Language