Table of Contents
Translation Model Setup
SelectTranslate offers free translation models, advanced models included with a subscription, and models you can connect with your own API key. You can also adjust model settings and prompts where supported.
Ways to access translation models
- Free models: Use the free models provided by SelectTranslate without a subscription or an API key.
- Subscription models: Subscribe to a plan to use its advanced models without setting up your own API key.
- Your own API key: Get a key from the model provider and add it to SelectTranslate to use the provider’s models.
Follow the steps below if you want to use an API key from your own account with a model provider.
Set up your own API key
- Open the model provider’s website and create an API key.
- Open the SelectTranslate extension panel. At the bottom, choose Settings → Translation Model, then add your provider under Custom Model.
- Enter your API Key and select or enter a Model Name. Use the default API endpoint unless your provider tells you to change it.
- Click Test. If the test fails, follow the error message and check the API key, model name, and endpoint.
- When the test passes, click Save. Return to the extension panel and select the model to start translating.

Make sure your provider account has access to the model and enough quota to use it. Your provider bills you for usage through your API key. Keep the key private.
Adjust model parameters and prompts
Go to Settings → Translation Model and open the model you want to adjust. Settings vary by model and are available for supported subscription models and models connected with an API key. Traditional machine translation engines and some free models don’t offer these controls.
| Setting | What it controls |
|---|---|
| Maximum text length per request | Controls the amount of text included in a single translation request. Setting it too large may increase response time or exceed the input limits of the model. |
| Maximum number of paragraphs per request | Controls the number of paragraphs processed in a single request, affecting the text batch processing method and response efficiency. |
| Maximum number of requests per second | Control the number of translation requests initiated per second to avoid triggering the service provider's rate limit due to too frequent requests. |
| Temperature | Adjusts the randomness of the model's output. Lower values generally make the output more stable, depending on the model. |
| System Prompt | Set the model's translation role, presentation style, and ground rules. |
| User Prompt | Specify translation tasks, original text input and translation output format requirements. |
The defaults work for most translations. To change the style, edit the prompts while keeping the required variables and output format in the User Prompt. Test the same paragraph before and after your change. To undo your edits, click Restore Default Configuration.

