Correct. People often say open models are six months behind the frontier. But “open” is too broad, - a model that needs a cluster is downloadable, not but not local to the everyday person. I want a scaling law for the best model that fits on one 24GB GPU. On Artificial Analysis’ current index, GPT-5 scored 23 in August 2025. Qwen3.5-27B reached 23 roughly 6.6 months later. GPT-5.3 Codex scored 33 in February 2026; Qwen3.8-27B reached 34 about 6.3 months later. A composite score is not task by task equivalence of course. But the pattern has now repeated across two Qwen generations. im seeing people post about 'Qwen Law' - basically matching frontier capabilities about two quarters later. Now will this hold for Astra and Fable 5.1? Thats yet to be seen.