Recent model releases from Anthropic, OpenAI, and Google have driven rapid gains on Code Arena leaderboards, with Claude Opus 5.5 posting Elo scores above 1800 and Gemini 4 Argon and GPT-6 variants close behind on coding-specific evaluations involving web development, agentic workflows, and multi-step debugging. These advances stem from expanded context handling, native tool use, and iterative fine-tuning on real engineering tasks, shifting competitive positioning as labs integrate the models into platforms like Copilot and Codex. Trader sentiment reflects this momentum alongside typical year-end release cycles, though product timelines can slip and benchmark criteria emphasize human preference votes that reward demonstrated capabilities over marketing claims. Key upcoming catalysts include any additional frontier updates or tool integrations before December 31.
如果 Arena.AI 排行榜(arena.ai/leaderboard/text)上的任何模型,在 2026年12月31日、下午11:59 ET 前於「Coding」的「Leaderboard」分頁達到至少指定的 Arena Score,本市場將判定為「是」。否則,此市場將以「否」結算。 此市場將根據 https://arena.ai/leaderboard/text/coding-no-style-control 上「Text Arena | Coding」排行榜分頁中「Score」欄的結果進行判定,且樣式控制功能須關閉。 本市場的結算來源為 arena.ai/leaderboard/text 上的 Chatbot Arena LLM 排行榜。如果檢查時無法使用此結算來源,本市場將保持開放,直到排行榜重新上線,並以重新上線後的首次檢查結果結算。若該來源永久無法使用,本市場將結算為「否」。
機器翻譯。如有爭議,以原文為準。