Current foundation
Execute faithfully
How can researcher-authored equations become inspectable execution plans with reproducible results across runtimes?
Open computational neuroscience
From equations to verifiable neural dynamics.
Next Brain connects Brian2 models to Rust, GPU, browser and distributed execution—then uses that stack to ask which neural details must remain explicit.
Live system map
Model → evidence
- 01ModelBrian2 equations
- 02PlanB2IR identity
- 03ExecuteCPU · GPU · MPI
- 04EvidenceAuditable bundle
Current evidence
4.13M / 24.1B
neurons / recurrent synapses
Macaque multi-area engineering run
139,255
FlyWire neurons
15.1M weighted connectome edges
87.14%
MNIST test accuracy
fixed dynamics + external readout
5 paths
CPU · Metal · CUDA · WASM · MPI
explicitly bounded support
Engineering results and scientific interpretation are reported separately.
Independent applications
Two ways to enter the research.
Each application opens on its own route with its own runtime and assets—never inside an iframe.
Research map
One stack.
Four connected questions.
We organise the work by scientific question—not by repository, backend or demo.
Current frontier
Scale responsibly
How far can models travel across local GPUs, browsers and clusters without losing model identity, provenance or resource bounds?
Active research
Compress what we can
Which local dynamics can proxy states and procedural connectivity replace—and which motifs must remain explicit?
Exploratory
Explain what emerges
Can perception, adaptation and safety behaviour be isolated through interventions that separate capability from mechanism?
Neural Lab · browser sample
Don’t just read the claim.
Run a bounded experiment.
Adjust excitability and execute a deterministic 64-cell leaky integrate-and-fire network locally. This compact JavaScript sample demonstrates the public interface; research experiments use the verified WASM execution path.
64 cells · 500 ms · dt 1 ms · seed 783
neuron index
biological time · 0–500 ms
The sample has not run yet.
Evidence ledger
What we know.
What remains open.
Every public claim should carry its model identity, code version, environment, result and boundary.
ProgrammeCurrent evidenceBoundary / next test
Engineering verifiedRuntime substrate
A shared model identity now reaches native Rust, GPU, browser and MPI execution paths.
Broader model compatibility and independent replication remain open.
Boundary explicitMulti-area model
Full-scale 4.13M-neuron / 24.1B-synapse bounded engineering runs have completed.
Scientific equivalence, repeatability and speed / cost advantage are not yet closed.
Scientific evidenceFlyWire vision
A fixed real-connectome network carries task information readable by an external classifier.
Matched random rewiring, multiple full seeds and internal plasticity are still needed.
Boundary foundApproximate dynamics
Richer proxies and selective retention expose a measurable fidelity–cost frontier.
The E1 plateau and 393 Hz artefact motivate stateful proxies and automatic motif selection.
Long-term blueprint
The roadmap is a sequence of testable questions—not a promise of inevitable scale.
- 01Verified execution substrateModel identity · portable plans · bounded runtimes
- 02Large-scale explicit modelsIndependent references · multiple seeds · cost accounting
- 03Selective structureProcedural connectivity · motif retention · fidelity frontier
- 04Stateful local proxiesShort memory · adaptive dynamics · boundary recovery
- 05Cross-scale memoryPersistent state · route-based reactivation · continual adaptation
Contact
Research questions are welcome.
Replication work, technical collaboration and careful discussion of methods or evidence.