OpenAI’s GPT-6 Astra series currently anchors second place on the arena.ai Code Arena WebDev leaderboard behind Anthropic’s Claude Opus 5.5 and Sonnet 5.5 models, driving the 81.5% market-implied odds for OpenAI as the second-best lab by October 31. Recent September releases of GPT-6 Astra and Claude Opus 5.5 shifted the frontier, with Anthropic models holding higher arena scores in agentic front-end tasks while OpenAI leads in specific categories such as data analytics and content tools. Google, Alibaba, and Meta trail by 100+ points, keeping their odds in the low single digits. With the October 31 snapshot less than three weeks away, any new model drops or sharp gains in community votes could still alter the lab ranking before resolution.