Compiler Benchmark Suite · CVM-1+CVM-2
Compiler Performance
Cross-domain benchmark rollup from the Organizational Intent Compiler proof pack. Intent Fidelity, coverage, projection accuracy, and Trust Lock stability across Healthcare, Enterprise, and Financial governance dialects.
Last generated: Jul 7, 2026
Benchmark table
| Benchmark | Intent Fidelity | Coverage | Projection Accuracy | Trust Lock |
|---|---|---|---|---|
| Healthcare | 88% | 100% | 96% | ✓ Verified |
| Enterprise | 88% | 100% | 95% | ✓ Verified |
| Financial | 88% | 100% | 97% | ✓ Verified |
Benchmark health
Healthcare PASSEnterprise PASSFinancial PASSCanonical IR PASSProjection PASSTrust Lock PASSDiff Stability PASSValidation Engine PASSBinding Graph PASS
Trust alignment (C4)
CVM-2 synthetic alignment scenarios — Intent Binding Graph + Trust Validation Engine.
Validation Fidelity
100%
Runtime Alignment
100%
Evidence Completeness
100%
Intent Drift (aggregate)
Low
Scenarios: 4/4 PASSOpen Trust Alignment →
Evolution fidelity (C3)
Healthcare v1 → v2 intent evolution benchmark — expected vs observed clause changes.
| Metric | Expected | Observed |
|---|---|---|
| Evolution Fidelity | 100% | 100% |
| Added clauses | 3 | 3 |
| Modified clauses | 1 | 1 |
| Removed clauses | 0 | 0 |
| Unchanged clauses | 2 | 2 |
Trust impact: 1 high · 2 medium · 1 lowOpen Compare Frameworks →
Compiler benchmark history
CVM-1+CVM-2
v0.1 — current release
Healthcare IF88%
Enterprise IF88%
Financial IF88%
Proof pack & compiler
Full CVM-1 artifacts, per-domain contracts, and reproduction steps.