Claude Opus 5.5, released September 22, 2026, as a hybrid reasoning large language model optimized for agentic coding and knowledge work, currently scores 67.9% on the newly launched HLE-Diamond benchmark—trailing only Google’s Gemini 4 at 69.1%. This close positioning, combined with Anthropic’s accelerated release cadence from Opus 5 in July, drives trader consensus around strong implied probabilities for the next Opus (5.6+) achieving 55%+ or 60%+ on the refined 1,000-question Humanity’s Last Exam variant. Competitive pressure from OpenAI’s GPT-6 series and Meta models, plus evaluation costs at higher effort settings, keep outcomes fluid while no official 5.6 timeline has emerged.