Skip to content
Open Source

Open research, open code

VERIAUDIT aims to release its evaluation code, dataset, benchmark methodology, and scoring framework openly, so the core measurement can be independently reproduced and scrutinized.

Open source

  • Evaluation framework (EquivEngine, GapEvaluator, model adapter interfaces)
  • Released dataset and schema, with the 200-scenario VeriAudit-X benchmark to follow after human verification
  • CLSG benchmark methodology and scoring framework
  • Reproducibility documentation (environment, prompts, evaluation configs)

Commercial / hosted

  • Hosted, large-scale evaluation runs
  • Proprietary / closed-source model testing via API
  • Automated continuous monitoring
  • Custom enterprise audits and governance report packages