ZUNA

ZUNA is a 380M-parameter BCI foundation model for EEG data, a significant milestone in the development of noninvasive thought-to-text. ZUNA reconstructs, denoises, and upsamples EEG data across arbitrary channel layouts and is built for researchers, clinicians, and BCI developers using real world data. Zyphra is excited to announce ZUNA, our first foundation model trained on brain data. We believe thought-to-text will be the next major modality beyond language, audio, and vision enabled by noninvasive brain–computer interfaces (BCIs). ZUNA is an early effort to build general foundation models of neural signals that can be used to understand and decode brain states. ZUNA is a key component in our mission to build human-aligned superintelligence. Over time, we see these models forming the foundation of thought-to-text agentic systems. ZUNA is a 380M-parameter diffusion autoencoder trained to denoise, reconstruct, and upsample scalp-EEG signals. Given a subset of EEG channels, ZUNA can: E

Zyphra Back Model Feb 18, 2026 San Francisco, California ZUNA: BCI Foundation Model Advancing Towards Thought-to-Text ZUNA: BCI Foundation Model Advancing Towards Thought-to-Text ZUNA is a 380M-parameter BCI foundation model for EEG data, a significant milestone in the development of noninvasive thought-to-text. ZUNA reconstructs, denoises, and upsamples EEG data across arbitrary channel layouts and is built for researchers, clinicians, and BCI developers using real world data. Chris Warner, Jonas Mago, Jonathan Huml, Beren Millidge Read Technical Report Hugging Face GitHub No headings found o

related reading

• How we collected 10,000 hours of neuro-language data in our basement - Conduitcondu.it
• [2211.02625] MAEEG: Masked Auto-encoder for EEG Representation Learningarxiv.org
• POYO-1poyo-brain.github.io
• Generalist - GEN-0 / Embodied Foundation Models That Scale with Physical Interactiongeneralistai.com
• Brain–computer interface - Wikipediaen.wikipedia.org
• Frontiers | Decoding EEG Brain Activity for Multi-Modal Natural Language Processingfrontiersin.org
• Neuronpedianeuronpedia.org
• Towards a “universal translator” for neural dynamics at single-cell, single-spike resolutionarxiv.org
• On the Opportunities and Risks of Foundation Modelsarxiv.org
• Frontiers | Feature Extraction and Classification Methods for Hybrid fNIRS-EEG Brain-Computer Interfacesfrontiersin.org
• How Many EEG Channels Should I Use In My Research? [Less Than You Think!] | Mentalabmentalab.com
• Neural Data Transformerarxiv.org