We’re delighted to announce version 1.2.0 of BabelFishAI, which adds Google Gemini as a new provider for both dictation and text!
Gemini 3.5 Transcribe for dictation
Your dictation is transcribed by Gemini 3.5 Transcribe, Google’s transcription model, through its Interactions API. The extension asks Google not to save your dictation in this API’s history.
Gemini 3.8 Flash for text
For rephrasing, correction, and translation, you can choose between two models:
- Gemini 3.8 Flash: the default model, with reduced thinking to respond in one to two seconds
- Gemini 3.5 Flash-Lite: the fastest
Your API key, your terms
As with the other providers, you use your own API key, which you can create on Google AI Studio. Prefer a key from a project with billing enabled. With a free key, Google may use your content to improve its products, potentially including human review, unless you are in the European Economic Area, Switzerland, or the United Kingdom. A note in the Gemini panel in the options reminds you of this.
Still free to combine providers
Gemini joins Mistral AI, OpenAI, and compatible servers (Custom/LiteLLM). For example, you can dictate with Gemini and translate with Mistral AI, or vice versa, depending on your preferences and API keys.
Also in this release
- Firefox 140 or later: during installation and updates, Firefox displays the data the extension sends to the provider you configure.
- Recording badge: a red dot in the extension’s icon during recording.
- Visited pages: the margin the extension adds at the top of the page disappears as soon as its banner closes, and the language declared by the page is no longer modified.
- No new permissions.
Updating
The extension updates automatically through the Chrome Web Store and Firefox Add-ons. Remember to update it on all your devices: versions earlier than 1.2.0 do not recognize Gemini. Then head to the options to configure your Gemini API key!