Echo finds Claude Mythos 5.1 GPT-5.6 Sol GLM-5.3 mostly disagree on vulnerabilities
Research by software security firm Echo has found that three frontier AI models—Claude Mythos 5.1, GPT-5.6 Sol, and GLM-5.3—agreed on only a single vulnerability out of 121 potential code flaws when auditing the same open-source software. Echo reported that 83% of the identified issues were flagged by a single model alone, highlighting significant discrepancies in overlap, severity ratings, validation passes, and costs across the tools. Open article →