Build a Model
A practical path from a simulation idea to a reviewable Spawn Model.
Start with the question
A useful Model begins with a clear system, not a token. Define what changes over time, what participants can influence, and what result the Model is meant to produce or explore.
Minimum submission
A Model submitted to Spawn should include:
- a plain-language purpose and intended audience;
- complete state, input, action, and output definitions;
- transition rules and timing;
- randomness and seed behavior, if used;
- starting-state construction;
- data sources, licenses, transformations, and limitations;
- deterministic serialization and checkpoint rules where replay is claimed;
- tests for normal, boundary, invalid, and interrupted execution;
- a benchmark showing compute, memory, storage, and expected runtime cost;
- a license and proposed Designer royalty policy.
Build process
IDEA → SPECIFICATION → IMPLEMENTATION → TESTING → BENCHMARK → BLUEPRINT REVIEW
Spawn Labs can assist with research, specification, code, testing, documentation, and visualization. AI assistance does not replace human review of scientific claims, security, economics, or legal requirements.
Before publication
Confirm that another qualified person can understand the Model, reproduce its claimed behavior, identify its assumptions, and estimate the cost of running it. Publishing a Model does not automatically make it launchable or guarantee adoption.