Resolution is based on the chatbot arena LLM leaderboard (https://lmarena.ai), specifically the company with the highest Arena Score in the Overall category, without style control or show deprecated, at the end of September 30th, 2025 11:59PM ET.
In the case of a tie, all companies tied for 1st place resolve to equal probability, such that they sum to 100%.
See also:
/Bayesian/which-company-has-the-best-ai-model
/Bayesian/who-will-have-the-best-texttoimage-SO0uN6suuS
/Bayesian/who-will-have-the-best-texttovideo-AtZ0CdIc8Z
/Bayesian/which-company-has-best-ai-computer
/Bayesian/which-company-has-best-vision-ai-en
/Bayesian/which-company-has-best-search-ai-mo
Previous months:
/Bayesian/which-company-has-best-ai-model-end
/Bayesian/which-company-has-best-ai-model-end-I0QsydsZuz
/Bayesian/which-company-has-best-ai-model-end-0CRdhqptRl
1,000🏅 Top traders
| # | Name | Total profit |
|---|---|---|
| 1 | Ṁ2,082 | |
| 2 | Ṁ1,160 | |
| 3 | Ṁ793 | |
| 4 | Ṁ661 | |
| 5 | Ṁ286 |
People are also trading
@Dulaman yeah they gamed it when it mattered, then lmarena made style control default which made it much harder to game, so ppl stopped gaming it for preferences (since that didn't even get you headlines anymore) so 2.5 pro is eternal slopking
@Bayesian Google's in trouble. Its Gemini 2.5 model has just been dethroned from its #1 position by... Gemini 3
@JoshSnider neat! i'd be surprised if they released opus 4.2 before a new sonnet version (oh ig they did increase its context window to 1m, maybe that counts enough to make it not surprising)
