Arena still ranks Google's Gemini 4 Argon first, with year-end bets split

Arena’s latest official text leaderboard still ranks Google’s Gemini 4 Argon first. That confirms the lead reported earlier, rather than marking a new change. Traders’ expectations for which company will have the top AI model at year-end remain close, with Google slightly ahead of Anthropic. Arena’s separate Agent leaderboard showed Argon in seventh place, but that is a different ranking from the text leaderboard relevant to this race.

The latest notable shift in traders’ expectations came on October 5, when Google’s chances rose from unlikely to a toss-up. Bets had eased on October 2, recovered a little the next day and slipped again on October 4 before the jump. Arena’s October 2 text ranking, which placed Argon first, was the most plausible reason for the move, according to Arena’s leaderboard changelog and rankings. The listing is an early signal, not a settled lead: Argon’s preliminary score of 1,525 was based on 4,932 votes, far fewer than some established rivals.

Google announced Argon on September 30, describing it as a frontier model for coding, professional work and cybersecurity. Its first release was limited to trusted cyber defenders while safety testing continued. Google said broader access would come to paid API customers and Google AI Ultra subscribers, but gave no date. Earlier coverage noted that Argon’s top ranking had persisted even as uncertainty remained about whether it would last; the latest listing reinforces that status without resolving the question.

The outcome turns on which company tops Arena’s text leaderboard at year-end, not on a model announcement or a separate benchmark result. More user ratings could strengthen or weaken Argon’s position, while new models gaining enough votes—or a rival moving ahead—could change the race. Google has said wider access is planned, but has not announced when it will begin.