Framework / Providers
Choose where model work runs
Provider-backed nodes can use managed NeoSyntropy inference or an explicitly selected provider while the graph contract stays stable.Provider selection
- Set provider on a model-backed node when that capability needs a specific configured model path.
- Use NeoSyntropyProvider for framework-neutral chat gateway calls.
- Use a CallableProvider for a locally controlled implementation that satisfies the provider contract.
- Keep provider credentials in project configuration or server environment, never in browser code.
Sync and async execution
Pythonpython
result = workflow.run(request, state={}, client=client, tools=registry)
# In an async service:
result = await workflow.arun(request, state={}, client=client, tools=registry)