Live readingSimulated
Reads the mind's state, live
Attention, load, fatigue, emotion and intent, second by second.
Simulated signal, running live.
The Large Brain Model. AI trained on what we experience.
The leading AI model built on brain activity. It reads what we don’t say, from the wearables we already wear.
AI learned to talk by reading what we write. This model learns from what we experience: over 505,000 hours of brain activity.
The biggest brain dataset in AI. By a long way.
15 years, 4,000+ institutions. It can't be bought or copied.
150,000+ peopleHealthy, not hospital patients.
156 countriesNot just the West.
One methodRecorded the same way everywhere.
Training hours per model. The LBM's include 314,000+ hours of Emotiv's own recordings.
Why there's no GPT for the brain yet.
The hardware is ready. The data isn't.
Every headset differs
2 to 256 sensors, no common standard.
Every model does one job
Sleep, emotion, movement: each built from zero.
Every brain differs
Mood, sleep and age all shift the signal.
Cleanup eats the time
60–80% of research time goes on cleaning data.
The LBM learns from all of it at once. GPT, for brain signals.
What the model does. Out of the box.
Works on any headset
From 2 to 32 sensors, without retraining for each device.
No setup for each person
Learned from 150,000+ people. Works on day one.
Small enough for an earbud
Beats models 8× its size. Can run on the device, so brain data never has to leave it.
Copes with real life
A sensor slips? It fills the gap.
Sees change over months
Tracks how a person's baseline shifts over months.
Illustrations, from simulated signal.
Why now.
AI has read the internet. The next data comes from sensors.
Neurotech money is surging
Record investment in brain wearables. Each needs a brain model.
36 brain tasks94 datasets14 model designs
The field has a benchmark
Meta's NeuralBench (2026): big models already win most tasks. The LBM enters after full training.
- Heart rate Taken
- Sleep Taken
- Steps Taken
- The brain Open
Platforms want the brain
Body data already feeds AI health apps. The brain is the missing layer.
Illustrations: the bars show the trend, not amounts, and the lit tasks are not exact results.
One model. Many products.
Focus earbudsNotice drifting attention and pause notifications.
Workplace safetySpot drowsiness and overload in operators, in real time.
Brain health checksStandard markers for screening and long-term studies.
Learning that adaptsSlow the lesson when a student is overloaded.
Games and XRIntent and engagement as a new kind of controller.
Research without a labPredict how brains respond to a product or ad, in hours.
A base for scienceA model researchers fine-tune instead of rebuilding.
AI that senses youBrain signal alongside vision, language and health data.
Built on published research.
Five peer-reviewed papers.
EEG2Rep
Hides parts of the signal and learns to predict them, without labels and despite noise.
SAMBA
Reads long stretches of signal at once. Best across 13 datasets.
EEGM2
Long recordings, on a model small enough for wearables.
EEG-X
Works on headsets it has never seen, whatever the sensor layout.
SpellerSSL
Typing with brain signal, with little setup. 94% of letters read right.
Four ways in. One model underneath.
Put it inside your hardware
Runs on the device itself, even offline.
Device makersPer unit shipped
License the platform
For pharma, research and enterprise. On-site option.
Platforms and enterpriseAnnual licence
POST /v1/decode { "device": "earbud" } → response { "state": "attention", "score": 0.92 }
Call one API
Signal in, brain states out.
Developers and researchersPay as you go
- My focus app
- Research study
- Share raw signal
Use your own brain data
Your readings, your apps, your call.
PeopleDirect access
“No general brain model exists because the data to train one has never existed. Until now.”
Questions, answered.
Want the full technical detail? Read the docs or email api@dcode.ai.
What is a Large Brain Model?
Like the models behind ChatGPT, but trained on brain signal (EEG) instead of text. It turns wearable signal into readings apps can use.
What hardware does it work with?
Any compatible EEG device, from 2-sensor earbuds to 32-sensor caps.
Does each person have to calibrate it?
No. It learned from 150,000+ people, so it works straight away.
What happens to people's brain data?
It can run on the device, so raw signal never has to leave it. The data belongs to the user.
Why can't someone else build this?
The data took 15 years to gather, from healthy people in 156 countries. Most datasets are hospital patients in a few Western countries.
How do partners get started?
Scope call, a test on your hardware, integration, then a licence.
Build on the Large Brain Model.
Get early access to the API.