Select the runtime and backend explicitly | Capability checks fail closed

Set up and run focused checks | Native code and generated wrappers | Numerical and review expectations | Keep data private and builds reproducible

From TABLO text to a model runtime | Solver and native boundaries | Numerical acceptance, diagnostics, and cleanup | Validation ownership

Estimate before solving | Set a memory budget | Request compact output and detailed diagnostics

Run the legacy workflow | Opt in to sparse execution | Package identity migration

Prepare a small model | Apply a shock and solve | Inspect the result

Prepare an external benchmark environment | Run the matched A/B workflow | Preserve the acceptance evidence