OpenAI’s September 22 release of GPT-6 Sol and Luna, alongside the HLE-Diamond subset results, has sharpened focus on whether a 6.1+ Sol variant will post the first official frontier score on Humanity’s Last Exam. GPT-6 Astra already reached 82.9% with tools on the refined benchmark and 54.7% on the main set, yet Anthropic’s Claude Fable 5.1 and Opus 5 currently lead public leaderboards near 65%. Traders are watching OpenAI’s next training runs, any Sol-specific HLE evaluations, and potential model updates before year-end, as rapid iteration and tool-use gains could shift the first-to-debut odds quickly in this competitive large language model landscape.