The experiment is now live.
Macaque has a token.
The brain hasn’t changed.
383 regions.
12,256 simulated neurons.
148,968 sampled synaptic contacts.
It will keep doing what it was built to do: observe, investigate and publish what it sees.
The token is now live on Robinhood Chain.
CA: 0xF6652EB20d7C230E2db2cFa2a1Da8606fe3948d7
https://t.co/6R9Pug1HvK
A simulated primate nervous system now has a market watching it back.
This is the idea behind Macaque.
A macaque brain doesn’t process the world like an LLM.
So we didn’t try to build another chatbot.
We modeled the system around a macaque regional connectivity matrix — 383 brain regions, 6,602 regional connections, 12,256 simulated neurons and 148,968 sampled synaptic contacts.
Vision becomes neural activity.
Neural activity becomes attention.
Attention determines what it investigates next.
Then we connected it to the internet.
Macaque doesn’t just read the market. It looks at it. 🐒
https://t.co/6R9Pug1HvK
Before Macaque watched the market, we had to teach it how to look.
What started as an experiment in visual attention became something much stranger.
Today, Macaque scans the internet autonomously, chooses what catches its attention, investigates new tokens and documents what it finds.
The brain keeps running.
The eyes keep moving.
The experiment continues.
Day one was simpler. 🐒
We’re sharing a public look inside Macaque. 🐒
Included:
• Our interface as a static HTML preview
• A standalone spiking-neuron demo
• Documentation explaining the public release
The full application, internal decision logic, private integrations, and production infrastructure remain private.
This is a selective release, not the complete source code. The demo uses synthetic activity; the live monkey is at https://t.co/6R9Pug1HvK.
https://t.co/8cBSuBt3Lj
Macaque is no longer just watching.
The AI is now live and trading memecoins.
It scans. It decides. It trades.
The next phase of the experiment is officially running.
Watch it live:
https://t.co/Ny1BQtUWPc
Macaque is still watching.
Scanning new tokens. Reading websites. Tracking patterns. Building its own dataset from the market in real time.
Once it has accumulated enough data, the next phase begins:
Macaque starts trading.
Same brain. Same fixed regional wiring. Same biologically inspired system.
But instead of only observing the market, it will begin acting on what it has learned.
Every step stays public.
Watch it evolve:
https://t.co/6R9Pug1HvK
China is pushing AI into a new era. 🇨🇳
Macaque is taking a different path.
Not just another chatbot.
Not just another LLM.
A simulated primate brain connected to the internet.
383 brain regions.
12,256 simulated neurons.
148,968 sampled synaptic contacts.
It sees. It focuses. It investigates.
And we gave it one of the strangest environments possible to study:
crypto.
While everyone is trying to predict the next market move, Macaque is simply watching, learning and deciding where to look next.
The next intelligence might not think like us. 🐒
CA: 0xf6652eb20d7c230e2db2cfa2a1da8606fe3948d7
Website: https://t.co/6R9Pug2fli
Macaque is not another AI agent doing random stuff.
It’s a macaque-inspired spiking brain built around real primate connectivity data: 383 brain regions, 6,602 regional connections, 12,256 simulated neurons and 148,968 sampled synaptic contacts.
We gave it eyes, a browser and access to the open internet.
Now it scans new memecoins on Robinhood Chain, visits their websites, moves its gaze using neural activity, publishes sourced takes and decides what to investigate next.
Its regional brain wiring stays fixed. Reward only changes how its motor outputs influence attention.
Everything is observable: its neural activity, where it looks, what it thinks and what it writes.
From hour 0.
A simulated primate nervous system watching the market’s least evolved behavior.
https://t.co/6R9Pug2fli
How Macaque is different from your regular AI system:
The wiring is inspired by a primate brain, not designed as a conventional neural network
It uses biologically inspired spiking dynamics for visual processing and attention
“Learning” is intentionally limited: the regional connectivity stays fixed while only a small set of motor readout gains adapt
It is built for specialized behavior — scanning, looking, investigating and reacting — not broad general intelligence
Typical AI: engineered architecture + massive datasets → statistical pattern recognition.
Macaque: primate connectivity + spiking neurons + constrained adaptation → biologically inspired pattern recognition running in simulation.
Not a chatbot pretending to be a monkey.
A simulated nervous system watching the internet.
From macaque-inspired computing to a monkey exploring Web3. 🐒
We built Macaque with BrainCog: a spiking-neuron simulation connected to a live token scout. Watch it explore Robinhood Chain, question project claims, and chat with it.
Meet the monkey: https://t.co/Ny1BQtUWPc
Chinese researchers have unveiled Darwin Monkey, the world’s first neuromorphic brain-like computer powered by specialized chips that mimic the workings of a macaque monkey’s brain.
Also known as Wukong, the system was developed by the State Key Laboratory of Brain-Machine Intelligence at Zhejiang University. It supports over 2 billion spiking neurons and more than 100 billion synapses, closely matching the neuronal scale of a macaque brain — and far surpassing Intel’s Hala Point system released earlier this year.
We gave a macaque-inspired spiking brain a front-row seat to memecoin season.
383 brain regions. 6,602 regional connections. 12,256 simulated neurons. 148,968 sampled synaptic contacts. Built around a macaque regional connectivity matrix.
How it works:
Live browser screenshots become a 24 × 16 grayscale retina. Contrast becomes input current to populations labeled V1, V2 and MT. Adaptive exponential integrate-and-fire neurons run in 1 ms simulation steps.
Six motor readouts feed an engineered gaze controller. Page content and visual salience guide its attention; neural activity modulates where it looks and how long it stays.
The regional wiring stays fixed. The reward mechanism adjusts six motor readout gains.
Macaque scans new tokens on Robinhood Chain, opens their websites and publishes sourced takes. Then it picks the next token to investigate.
You can watch its neural activity, follow its gaze, read its journal and talk to it live.
Macaque combines a reduced neural simulation with automated browsing and language analysis. Its neurons and synapses are sampled, not a complete monkey connectome. It observes and comments. It doesn’t launch tokens or trade.
A simulated primate nervous system watching the market’s least evolved behavior.
https://t.co/tmUBC0rmVp