Recent model releases have intensified competition for the #2 lab position on arena.ai’s text leaderboard with style control enabled, producing near-even implied probabilities among Anthropic, Google, and Meta. Google’s Gemini 4 Carbon launch on October 28 vaulted it to the top overall in several benchmarks, while Anthropic’s Claude Opus 5.5 and Meta’s Muse Wonder Full remain within a few Elo points on style-controlled evaluations, reflecting strong reasoning and instruction-following capabilities. OpenAI’s GPT-6 Astra and GPT-6.1 variants trail in this specific metric despite high raw scores elsewhere. Chinese labs such as Moonshot and DeepSeek continue incremental gains but lack the frontier consistency needed to close the gap before year-end. Traders are watching further releases and benchmark updates through December for any decisive shifts in lab rankings.