Learn how to integrate Gemini with POEditor to translate, refine and evaluate translations with AI in your localization projects.

AI translation and evaluation features are available on all POEditor subscription plans, including the Free plan.

Setting up the Gemini integration

To connect Gemini, open the Integrations page in POEditor and choose Gemini.

Enter your Gemini API key, check the box confirming that you understand AI features are experimental, select the model you want to use, then save your settings.

If you plan to use Gemini as your primary AI provider, you can set it as the default. It will then be used automatically whenever you run AI translation, AI adjustment or Quality Evaluation, unless you choose another provider.

AI integrations are configured at account level. Project owners, admins and contributors can connect their own Gemini integration, or use one shared by the project owner, as explained in the following section.

Allowing contributors to use your Gemini integration

You can share your Gemini integration with contributors who are part of your localization projects.

Enable the option Contributors Can Use Your AI Integration in Account Settings. Contributors can then use your Gemini integration without providing their own AI API key.

AI Translation with Gemini

Once Gemini is configured, AI translation is available from any language page in your project.

To translate an individual string, click an empty translation field and select Translate with AI.

If you want to translate multiple strings in one go, use the AI menu at the top of the Language page.

Choose the source language and the target language(s). You can also use Advanced Settings to restrict translation to strings with specific tags. Only strings that don't already have translations in the selected languages will be processed.

AI post-editing via Gemini (adjustment prompts)

Gemini can be used to refine existing translations, whether they were generated with AI, imported, or added manually.

Once an AI translation is available, Gemini can create a different version according to your preferences. Built-in adjustment prompts include Formal, Simplify, Shorter and Rephrase, and you can also create your own adjustment prompts.

You can also update existing strings with the option Edit with AI.

When Quality Evaluation highlights issues, use the QE Fix adjustment prompt to generate an improved version of the translation.

AI Quality Evaluation using Gemini

Gemini can also power Quality Evaluation when selected as your AI provider.

Each translation is assigned a score between 0 and 100 using the MQM (Multidimensional Quality Metrics) framework. Any detected issues are grouped by category and severity level.

To use Quality Evaluation, your project must have a Default Reference Language configured, or you can choose another project language to serve as the reference.

When a glossary is assigned to the project, the relevant terms and definitions are automatically included in the evaluation process.

To start an evaluation, open a language page and select the desired scope:

• Unscored translations only - checks only translations that have not been evaluated yet.
• All translations - evaluates all translations again, including those with existing scores.

The Quality Evaluation score is shown next to each translation when you hover over it.

Select a score to see the issues detected during evaluation and their details.

After reviewing the results, use the dedicated feature to generate an improved translation.

AI prompt customization

The default AI prompts are designed to balance translation quality, performance and cost efficiency, but you can adjust them to better fit your localization workflow.

In your Gemini integration settings, select Add custom prompt to update the default translation prompts (Translate, Translate Plural and Translate Bulk), customize the built-in adjustment prompts (Formal, Simplify, Shorter, Rephrase and QE Fix), or add new adjustment prompts.

Translation prompts define how Gemini handles AI translations, while adjustment prompts are used to modify existing translations, including QE Fix for issues identified through Quality Evaluation.

When customizing prompts, make sure to include the required placeholders, as recommended in Tips for adjustment prompts.

Custom models

You can use a custom AI model trained with your own data in POEditor instead of the default models available in the integration.

Once Gemini is connected, your custom model will be available in the model selector.

Workflows: AI translation step

You can add a Translate with AI step to your Workflows to automate translations with Gemini as the selected AI provider.

Some workflow templates include this step by default, or you can add it when creating a Custom workflow.

Select the source language and target languages for the translation step. Source strings can come from Project Terms or an existing project language. If your project has a Default Reference Language, it will be chosen automatically.

You can use languages inherited from the previous workflow step, choose specific languages manually, or apply the step to all project languages. Advanced options let you filter which strings are sent for translation by tag.

Using the official POEditor MCP server

If you use an MCP-compatible AI assistant, you can connect it to POEditor through the official POEditor MCP server.

The MCP server allows you to handle localization tasks such as managing projects, updating translations, exporting files and running QA checks directly from your AI assistant, without the need to create API scripts or switch between tools.