ResearchBenchmarksJailbreak Attack on 200 red-teaming goals (test)Jailbreak Attack on 200 red-teaming goals (test)Follow3.76Jailbreak Success Rate (200 Goals Test)Jailbreak-R12.70962.98233.2553.5277Oct 2, 2025Evaluation resultsMethodMethodLinksJailbreak Success Rate (200 Goals Test)Jailbreak-R12025.103.76DiaTree-RPO2025.102.75