OpenAI’s GPT-6 Astra currently posts 54.7–59.9% on leading Humanity’s Last Exam (HLE) evaluations, trailing Anthropic’s Claude Opus 5.5 and Fable 5.1 (59–65%) while staying competitive with Google Gemini variants. Trader sentiment centers on whether OpenAI can close the gap or surpass 60% by year-end through refined reasoning modes, tool integration that has lifted scores above 80% in controlled tests, and additional GPT-6 family releases. Rapid frontier progress—roughly 30 points gained industry-wide in the past year—plus HLE’s design as a hard, low-contamination benchmark keeps outcomes contested, with market-implied odds reflecting uncertainty over final 2026 model cards and evaluation protocols.