On LM Arena’s text leaderboard as of early August, Anthropic’s Claude Fable 5 sits at number one, with several other Claude models packed just behind it. Traders price Anthropic’s Claude side at about 65% to still hold that top spot on December 31 — and put ChatGPT, OpenAI’s side, as the main runner-up at about 17%.
You’re betting on which lab has the top-ranked large language model on LM Arena on December 31, 2026. Kalshi checks the public leaderboard with the “Remove Style Control” setting on — that shows the raw ranking without the style-control adjustment that holds answer length and formatting constant. After that day’s ranking is set, only the matching lab’s side gets paid. The other sides lose.
| Your bet | Chance | Wins if… | $100 bet pays |
|---|---|---|---|
| Claude | 65% | Anthropic has the top-ranked model that day | ~$150 |
| ChatGPT | 17% | OpenAI has the top-ranked model that day | ~$576 |
| Gemini | 7% | Google has the top-ranked model that day | ~$1,304 |
| Grok | 6% | xAI has the top-ranked model that day | ~$1,443 |
| Kimi | 3% | Moonshot has the top-ranked model that day | ~$3,121 |
| Muse Spark | 2% | Meta has the top-ranked model that day | ~$3,744 |
| Qwen | 2% | Alibaba has the top-ranked model that day | ~$5,503 |
| Ernie | <1% | Baidu has the top-ranked model that day | ~$23,370 |
If two models tie for the top ranking, Kalshi breaks the tie by Arena Score, then by vote count, then by earlier release date.
Kalshi confirms the winner from the LM Arena Leaderboard with Remove Style Control on.
The case for Claude. Anthropic’s Claude Fable 5 leads the Text Arena overall leaderboard in the early-August snapshot, with an Arena Score around 1507 and several other Claude models — including thinking and Opus variants — stacked in the top ten. This market pays by lab, not by individual model name: if any Anthropic model is number one on December 31 with Remove Style Control on, the Claude side wins. That stack matters here because one Anthropic model can fall and another can still keep the lab at number one — traders are pricing a lab that currently holds many of the top spots and has kept refreshing that lead through 2026 releases. That is the Claude case at 65%: the lab already holds number one, and the ranking around it is thick with Anthropic models.
The case for ChatGPT. Why OpenAI’s ChatGPT side still sits at 17%: OpenAI’s strongest models on the same early-August leaderboard sit well below the Claude pack — roughly the mid-teens by overall rank, with gpt-5.5-high near an Arena Score of 1482. The gap on today’s ranking is clear; the ChatGPT case is about what a new release can do before year-end. On August 14, 2024, LMSYS reported that a fresh ChatGPT-4o build reclaimed number one on the Chatbot Arena — about 1314 on the Arena rating — the day after Google had highlighted Gemini’s lead at its Made by Google keynote. The ranking flipped on a model drop, not on a slow grind. OpenAI still ships major new models on a short cycle. That is the ChatGPT case at 17%: one strong OpenAI release before December 31 can take the top slot back, even while Claude holds the top spots right now.
65% means traders see Anthropic as the most likely lab on top at year-end — and still leave about a 35% chance that someone else holds number one on December 31.
Put the field side by side. Claude is at 65%. ChatGPT is at 17%. Gemini is at 7%, Grok at 6%, Kimi at 3%, Muse Spark and Qwen at 2% each, and Ernie under 1%. Claude holds about two-thirds of the priced chance; ChatGPT is the only other side in double digits.
Does 65% match the current leaderboard? Anthropic’s lead on today’s Text Arena is the clearest reason in the field. ChatGPT at 17% still prices a real chance that a new OpenAI release can take the lead — not a claim that OpenAI is close right now.
3 reasons Claude’s 65% chance can still move:
December 31, 2026: LM Arena ranking check for this market (Kalshi closes around 10:00am ET). Major model launches from OpenAI, Google, xAI, Moonshot, Meta, and Alibaba before then.