Anthropic’s recent frontier releases, including Claude Opus 5.5 and Sonnet 5.5 in late September plus the October 7 Haiku 5.5 launch, have driven its 68% implied probability as the #2 AI lab by month-end under style control benchmarks. These models lead multiple independent indices on reasoning, coding, and output consistency while introducing refined style controls that reduce mannered prose and improve natural tone. Google’s Gemini 4 Argon release on September 30 provides a credible alternative at 23.5% odds, though it trails on several composite scores. Meta sits at 8% amid slower frontier progress, while lower-probability labs face steep gaps in demonstrated capabilities and recent benchmark results. Traders are watching any final October updates that could alter the ordering before resolution.