Moshi, every mention
6 scenes (2026), the whole family · ← back to Moshi
tap a year for its mentions
every year 2026 anyone Neil Zeghidour 8
Verbatim, from the transcripts: the passages where Moshi comes up
Voice AI’s Big Moment: Top Researcher on Why Everything Is Changing (Neil Zeghidour, Gradium AI)
- ▶ 25:45 Neil Zeghidour But in six months, we have a team of, a core team of four people, and then six, we were able to ship a model trained from scratch called Moshi, that is still to the day the only full duplex model.
- ▶ 36:15 Neil Zeghidour Uh, the Moshi architecture, like pretty much every, uh, uh, model right now, even the Voxstral model that was released by Mistral, uh, two weeks ago is also, uh, based on, uh, on our, uh, framework.
- ▶ 46:01 Neil Zeghidour We, you know, we did Moshi a year and a half ago.
- ▶ 50:55 Neil Zeghidour Moshi, again, people can, can talk to it today. 2 times in the scene
- ▶ 57:24 Neil Zeghidour So what we did for mostly is we started from the text model and then we, we, we, we, 2 times in the scene
- ▶ 1:09:23 Neil Zeghidour We have an appendix in the Moshi paper around how we could break so easily any watermarking that was supposed to be state of the art.