Recent model releases have tightened the race on Code Arena WebDev, where community Elo ratings reward agentic frontend coding performance. Anthropic’s Claude Opus 5.5 Max currently leads at 1815, with OpenAI’s GPT-6 Astra Max a close second at 1788; traders assign OpenAI the highest implied odds for second place by year-end because its rapid iteration cycle and full-stack strengths could close the gap before December voting closes. Google’s Gemini 4 Argon sits third but trails by over 100 points, while Chinese labs like Alibaba’s Qwen and Moonshot’s Kimi remain competitive lower down. Key swing factors include upcoming releases, continued Pareto-frontier gains on cost-efficiency, and any late-year shifts in multi-step reasoning benchmarks that could reorder lab standings.