Speaker Recognition and Faster Language Models Available

Work more efficiently with improved transcription and Gemini 3.5 Flash.

Date: 2026-06-09

Category: News

New models are now available in AI-Public to help you work even more efficiently and accurately within the public sector. The main addition is the new model for the transcription module: gpt-4o-transcribe-diarize.

This model automatically recognizes when a different speaker is talking. This makes transcribing interviews, public participation meetings, or discussions with colleagues much more manageable. The text is immediately divided by person (for example, Speaker A and Speaker B), significantly reducing the time you spend manually editing conversation reports. The improved transcription with speaker recognition is applied automatically when you start the transcription module for your audio recordings.

In addition, we have added Gemini 3.5 Flash to the range of language models. This model is specifically designed for speed and efficiency. It is ideally suited for tasks where you need immediate results, such as generating short text suggestions for public communication, quickly summarizing incoming correspondence, or answering factual questions during your work.

With these updates, you can spend more time on the substance of your work, while AI-Public handles supporting tasks faster and more clearly for you.

The new models are now immediately available to all users within the organization.