Open Source

ZUNA "Thought-to-Text": a 380M-parameter BCI foundation model for EEG data (Apache 2.0)

An open-source foundation model reads EEG data to decode thoughts into text, released under Apache 2.0.

Deep Dive

Zyphra has released ZUNA, a 380-million-parameter foundation model designed for Brain-Computer Interface (BCI) applications. It processes EEG (electroencephalogram) data to translate neural activity into text. Released as open-source under the Apache 2.0 license, the model and code are available on Hugging Face and GitHub. This enables researchers to build non-invasive tools for thought-to-text communication and advance assistive technology.

Why It Matters

It accelerates non-invasive BCI research, paving the way for new communication aids and neurotechnology tools.