Recent releases from Anthropic and Google have accelerated gains on coding leaderboards such as LMArena Coding and WebDev Arena, with Claude Fable 5.2 and Opus 5.5 frequently posting the highest Elo ratings near 1800–1830 while Gemini 4 Carbon and Argon sit close behind at 1760–1560. These models demonstrate stronger performance on agentic tasks, multi-step code editing, and human-preference battles than prior generations, reflecting rapid iteration across labs. With only six weeks remaining until December 31, trader sentiment hinges on whether incremental fine-tunes or new frontier releases can push any system past the queried threshold before year-end, as historical patterns show top scores rising several dozen points per month amid intense competition.
This market will resolve to "Yes" if any model on the Arena.AI Leaderboard (arena.ai/leaderboard/text) reaches at least the specified Arena Score on the "Leaderboard" tab for "Coding" by December 31, 2026, 11:59 PM ET. Otherwise, this market will resolve to "No". Results from the "Score" column under the "Text Arena | Coding" Leaderboard tab at https://arena.ai/leaderboard/text/coding-no-style-control with style control off will be used to resolve this market. The resolution source for this market is the Chatbot Arena LLM Leaderboard found at arena.ai/leaderboard/text. If this resolution source is unavailable at check time, this market will remain open until the leaderboard comes back online and will resolve based on the first check after it becomes available. If permanently unavailable, this market will resolve to "No".