Show HN: GPT-3.5 answered, GPT-4 voided; I built an EOS gate for AGI (getswiftapi.com)

A new project has emerged that explores a concept termed "Void measurements" in advanced language models, notably GPT-3.5 and GPT-4. The study, released on September 23, 2026, highlights a distinct behavior in GPT-4, which returned zero-visible-byte responses in all test cases involving null queries, unlike GPT-3.5, which provided visible text consistently. This innovative experiment involves a cross-vendor analysis and presents a framework to interpret and categorize model outputs, assigning specific classifications such as "Void" to successful executions that yield no visible output, differentiating them from errors or failures. The significance of this research lies in its potential implications for the understanding and development of Artificial General Intelligence (AGI). By proposing a criterion for AGI that involves binding conditions across various domains, the authors seek to establish a nuanced approach to evaluating model behaviors. This publication not only redefines how we interpret outputs from leading AI models but also introduces a systematic method for verifying model execution through cryptographic attestations. Such advancements could pave the way for more robust, interpretable AI systems, enhancing safety measures and fostering confidence in AI deployment.