Anthropic’s September 2026 releases of Claude Opus 5.5 Max and Fable 5.1 Max have driven the Code Arena/WebDev leaderboard to new highs above 1815 Elo, establishing a clear lead over OpenAI’s GPT-6 Astra Max (1788–1789) and GPT-6.1 Sol Max (1758–1759). These gains stem from targeted improvements in agentic coding, multi-step tool use, and web development workflows, outpacing earlier 2026 models from Qwen, DeepSeek, and Google. With roughly 80 days remaining until December 31, trader sentiment hinges on whether labs can sustain incremental Elo gains through further scaling, specialized fine-tuning, or efficiency breakthroughs before year-end deadlines. Additional frontier releases, arena updates, or cost-optimized variants from OpenAI and Anthropic represent the key near-term catalysts that could shift implied probabilities on higher thresholds.