Researchers have developed a new method called NoTB to assess the correctness of register-transfer-level (RTL) designs generated by large language models (LLMs). Unlike existing approaches that rely on simulation or LLM-generated testbenches, NoTB uses cross-model formal consensus to infer correctness. The framework generates RTL implementations from multiple independently trained LLM families and applies Sequential Equivalence Checking (SEC) to identify provably equivalent designs. This approach allows for risk-coverage tradeoffs without requiring testbenches, providing a reliable basis for high-confidence triage. --- Why it matters: NoTB matters because it addresses the challenge of assessing functional correctness in early stages of LLM-generated RTL design. By enabling designers to make informed decisions about which designs are correct and which require further testing, NoTB can save time and resources in the development process. Source: https://arxiv.org/abs/2608.21962