Anthropic's AI Crown Is Safe Until June

Google I/O came and went without denting the market's conviction that Anthropic holds the best AI model.

Google I/O came and went without denting the market's conviction that Anthropic holds the best AI model.

Polymarket prices Anthropic at 99.9% to hold the top spot at the end of May, with Mistral the only alternative at 0.1%. The market has not moved in weeks. Even Google's annual developer conference, which ran May 14 to 15 and dominated AI coverage, failed to register a tick.

The strongest pre-conference signal came from Alex Heath at Axios, who reported on I/O day one that Google was "about to release" a new Gemini model. That was the one catalyst traders could have priced in. They didn't. The implication: whatever Google showed was not enough to challenge Claude Opus 4.6's position atop third-party rankings.

The broader AI sector is not short on momentum. Cerebras, the inference chip maker, surged 70% on its Nasdaq debut May 14 in the biggest IPO of 2026. Investor appetite for AI infrastructure is running hot. But hardware hype does not translate into a model quality challenge on a two-week timeline.

One wildcard remains. In April, reports surfaced that Sergey Brin is personally leading development of a Claude Code challenger inside Google. If that product ships and benchmarks above Anthropic before May 31, it would be the kind of black swan that reprices the market overnight. The 0.1% residual probability reflects how seriously traders take that scenario: not very.

The window is the real constraint. A challenger would need to launch, undergo independent evaluation, and be declared superior to Claude within roughly 13 days. Polymarket's "best model" criteria remain loosely defined, but no evaluation cycle moves that fast.

For now, 99.9% is less a statement about Anthropic's permanent superiority and more about the calendar. The race resets in June. Until then, the market has called it.