Data as of Jul 25, 2026 · Based on 2,718,867 AI responses across 9,511 prompts · See how Parse measures this
AVERIaveri.org · Low evidenceAVERI is a nonprofit focused on making third-party auditing of frontier AI effective and universal, ensuring rigorous safety and security assessments of powerful AI systems and their developers. It performs independent verification and evaluation of safety and security claims against relevant standards, using secure access to non-public information when needed. Its outputs include tools like BenchRisk for meta-evaluating benchmark risks, and it invites collaboration from researchers, policymakers, and AI developers.
Parse Score