René's URL Explorer Experiment

{"@context":"https://schema.org","@type":"WebPage","name":"AgentContract-Bench v2 — 293 Benchmark Scenarios","description":"AgentContract-Bench v2: 293 scenarios across 12 domains with 100% pass rate. Live LLM results from GPT-5.3, Claude Sonnet 4.6, and Mistral-Large-3.","url":"https://agentassert.com/benchmarks/","inLanguage":"en","isPartOf":{"@type":"WebSite","name":"agentAssert","url":"https://agentassert.com/"},"author":{"@type":"Person","name":"Varun Pratap Bhardwaj"}}

{"@context":"https://schema.org","@type":"Organization","name":"Qualixar","url":"https://qualixar.com","logo":"https://agentassert.com/brand/logo-mark.svg","sameAs":["https://www.linkedin.com/in/varun-pratap-bhardwaj-7ab63742/","https://x.com/varunPbhardwaj"],"founder":{"@type":"Person","name":"Varun Pratap Bhardwaj"}}

{"@context":"https://schema.org","@type":"SoftwareApplication","name":"agentAssert","applicationCategory":"DeveloperApplication","operatingSystem":"Any","description":"Agent Behavioral Contracts: Formal Specification and Runtime Enforcement for Autonomous AI Agents","author":{"@type":"Person","name":"Varun Pratap Bhardwaj"},"offers":{"@type":"Offer","price":"0","priceCurrency":"USD","description":"Patent Pending — Request Access"}}

{"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://agentassert.com/"},{"@type":"ListItem","position":2,"name":"Benchmarks","item":"https://agentassert.com/benchmarks"}]}

{"@context":"https://schema.org","@type":"FAQPage","mainEntity":[{"@type":"Question","name":"What is AgentContract-Bench v2?","acceptedAnswer":{"@type":"Answer","text":"AgentContract-Bench v2 is a comprehensive benchmark suite with 293 scenarios across 12 domains for evaluating AI agent behavioral contract compliance. It achieves 100% pass rate with the agentAssert framework."}},{"@type":"Question","name":"Which LLMs have been tested?","acceptedAnswer":{"@type":"Answer","text":"Live results are available for GPT-5.3 (Theta=0.688), Claude Sonnet 4.6 (Theta=0.823), and Mistral-Large-3 (Theta=0.813). All models were evaluated across the full 293-scenario suite."}},{"@type":"Question","name":"How are benchmark scores calculated?","acceptedAnswer":{"@type":"Answer","text":"The aggregate compliance score Theta is computed across all contract constraints (preconditions, invariants, guarantees, and recovery) weighted by domain-specific importance. A higher Theta indicates better behavioral contract adherence."}}]}