OpenAI’s GPT-5.6 Sol currently leads the company’s HLE results near 49.5% on Artificial Analysis snapshots, trailing Anthropic’s Claude Fable 5 and Opus 5 variants at 54.9–55.5%, while certain GPT-5.4 Pro configurations have posted up to 58.7% on provider-reported tool-augmented runs. Trader sentiment reflects rapid frontier progress on this 2,500-question expert-level benchmark—scores have climbed from single digits in early 2025 to the mid-40s or higher by mid-2026—driven by GPT-5 series scaling, chain-of-thought scaffolding, verification loops, and agentic tool use across math, science, and humanities domains. Anthropic’s edge stems from stronger multimodal reasoning and larger context handling, creating competitive pressure that favors continued OpenAI iteration through year-end releases. With four months remaining, market-implied odds price a strong chance of crossing 50% but assign lower probability to 60%+ or 65%+ thresholds, as further gains depend on unannounced model updates amid ongoing benchmark calibration challenges.