OpenAI’s September 22 release of GPT-6.1 Sol, positioned as a lower-cost model approaching GPT-6 Astra capabilities on agentic and professional tasks, has shaped trader views on whether the Sol line will first achieve a notable HLE debut. GPT-6.1 Sol posted mid-50s scores on HLE variants and HLE-Diamond subsets, trailing Anthropic’s leading Claude Fable 5.1 and Opus 5 models at 64-65% on the main benchmark while showing stronger tool-augmented results. Recent HLE-Diamond data highlighted OpenAI’s Astra variant reaching 82.9% with web and code tools, underscoring rapid capability gains but also the benchmark’s resistance to quick saturation. Upcoming catalysts include further OpenAI or competitor releases, HLE-Rolling updates, and new verified evaluations that could clarify Sol-series positioning relative to the frontier.