Recent OpenAI releases in the GPT-6 family, including GPT-6 Astra and GPT-6.1 Sol in September and October 2026, represent the main catalyst shaping trader views on the highest HLE score achievable this year. These models post competitive results on variants of Humanity’s Last Exam, a 2,500-question expert benchmark spanning science, math, and other fields, yet trail leading Claude Opus and Fable configurations from Anthropic on most public leaderboards, where top scores range from the mid-50s to low-70s percent depending on tool use and verification. Ongoing rapid iteration at OpenAI, combined with potential year-end model updates or refined evaluations, keeps the outcome open while underscoring how benchmark gaps between labs can shift quickly on new releases.