MANIFESTROMODELS
Systems built from our research.
Research, models and products are different things at Manifestro. A model listed here reflects its true state of readiness, we do not dress up experiments as finished technology.
Aether
IN DEVELOPMENTA speech-native AI model for real-time, full-duplex human interaction.
Aether is designed to listen and speak simultaneously, treating speech as a native modality rather than relying on the traditional speech-to-text → language model → text-to-speech pipeline. It is currently under active development and is not yet publicly available.
A model's status here is a promise about what it is, not what we hope it becomes.
Experimental means exactly that. Preview means a small group of partners is already stress-testing it. We move things forward one stage at a time.