Unreviewed preprint posted to GitHub. Makes quantitative cross-model claims: Claude Opus 4 93% convergence within 30 turns; GPT-4 71% within 30 turns (requiring 2.3 more turns per phase transition); PaLM 2 58% within 30 turns with more variable progression. No methodology section describes how GPT-4 or PaLM 2 data were obtained — no API access details, model versions, or experimental procedures for non-Anthropic models. Author uncredited; publication date estimated from IFLScience article (Jun 11, 2025) which cites it as the primary source for its cross-model replication claims. Quantitative Claude data mirrors Anthropic's system card; non-Anthropic figures are unsubstantiated.