Optimal Translation
You can choose Translation Models, Target Language and a Scoring model, and optionally enable Score by sentence and Scoring rationale as needed. After entering your text, click Translate to start translating. The results panel will automatically display the top-scored translation, and you can also manually select translations from other models.
Feature Overview
- Translation Models: The models used to translate the source text. It is recommended to select three or more models for translation.
- Target Language: The language into which the text will be translated.
- Scoring Model: The model used to evaluate and select the best translation. It’s recommended to choose powerful translation models such as DeepSeek, Gemini, or OpenAI to significantly improve selection accuracy. (Machine translation models should not be used as scoring models.)
- Score by Sentence: For a passage with multiple sentences, Score by Sentence evaluates each sentence individually. Click on a sentence to view its analysis and score. The results panel will automatically merge the highest-scoring translation for each sentence, and you can manually replace any sentence using the right-side Score by Sentence results. If you want to use the translation from a single model entirely, select that model from the Results dropdown.
- Scoring Rationale: When enabled, the scoring model analyzes and scores the translation, providing an explanation of the score (this will consume more translation credits). When disabled, the model simply scores the translations without analysis.
💡 The language used in Scoring Rationale matches the interface language you have selected.
🔔 If you encounter any issues or have suggestions, you can submit them on the Feedback page.
🔔 If you encounter any issues or have suggestions, you can submit them on the Feedback page.

