ModelManagementEN

🧠 What It Is

Model Management sets up the AI models for translation and AGENT.

Category	Select it when
Preset Models	You use common services such as DeepSeek, OpenAI, Anthropic, Google, ChatGPT, or SakuraLLM
Custom OpenAI Models	A service is compatible with the OpenAI API, for example a local model or a relay service
Custom OpenAI Responses Models	A service is compatible with the OpenAI Responses API
Custom Anthropic Models	A service is compatible with the Anthropic Claude API
Custom Google Models	A service is compatible with the Google Gemini API

🚀 Quick Setup

1. Select a preset model
2. Enter the API Key in Basic Settings
3. In Model Identifier, click Fetch and select a model from the list
4. Click Test

✅ Result: the test succeeds

💡 For an example, see [DeepSeek]

Models that use account sign-in

Some preset models (for example ChatGPT) use an account sign-in instead of a key:

1. Click Click to sign in
2. Sign in in the default browser. You can also use Copy link to open it in another browser

⚙️ Settings

Basic Settings

Setting	Description
Model Name	Used only for display in the app
API URL	Check whether the URL must end with /v1
API Key	You can enter more than one, one per line. They are used in rotation
Model Identifier	The model that is called. Click Fetch to read the list
Thinking Level	How deeply the model thinks. With Keep default, the parameter is not sent
Speed level	A fast mode of some providers. When it is active, it costs more

Task Settings

Setting	Description
Input Token Limit	Maximum input tokens for each translation task
Output Token Limit	Maximum output tokens for each translation task, 0 = automatic
Requests Per Minute Limit (RPM)	Maximum requests per minute, 0 = automatic
Concurrent Task Limit	Maximum tasks at the same time, 0 = automatic

Advanced Settings

Setting	Description
Context Window	AGENT only, 0 = automatic
Maximum Output Length	AGENT only, 0 = automatic
top_p temperature	Do not change if you are not sure. Wrong values cause bad results or errors
Custom Request Headers Custom Request Body	JSON format. Do not change if you are not sure

🗂️ Manage Models

Operation	Description
Duplicate	Copy the current model, for example to compare settings
Reset model	Restore the default settings
Delete model	Each category must keep at least one model

🔀 Where to Use Them

In the bottom action bar of the AGENT page, you can select the AGENT Model and the batch translation model separately. See [AGENT].