AI Models

Lexiflux use AI models to create AI Insights in Sidebar (you open it with blue binocular icon).

Also you can use AI for inline text translation. By default it uses Google Translate.

To configure Sidebar and inline translation see Dictionary & AI Insights Settings. There are separate settings for each language, so you can configure different settings for different languages.

AI Insights Types

AI Powered (Translate, Lexical, Origin, Explain, Sentence)

They use predefined prompts to generate insights. You can select AI model (ChatGPT, Claude etc) to run them.

Custom AI

With type "AI" you can define your own prompt for AI model.

Dictionary

There are a number of dictionaries embedded like Google Translate, Linguee Translator, MyMemory Translator, PONS Translator.

Site

You define what a URL to open and with which parameters. This is usefull for good translator that does not have API. Like glosbe.com.

Most of them should be open in external window, but some of them even could be open inside the Sidebar.

This is controlled by open in new window parameter.

AI models: OpenAI (ChatGPT), Google (Gemini), Anthropic (Claude) and others

To use commercial AI providers you need API KEY.

See AI Connections section how to get one for different AI providers.

Gemini 2.5 Flash is smart enough for even tricky language puzzles and can be used with free tier.

Ollama

You can install in the docker local AI Ollama model.

It is free, but requires about 4G RAM to run. Remember this is RAM for the Docker container, not for the host machine.

And it is not so smart as commercial models from OpenAI and Anthropic.