Next Claude Opus: Humanity’s Last Exam Debut?: how this market works

What you need to know

This market is asking how well the next major Claude AI model, specifically one labeled 'Opus', will score on a very hard academic test called Humanity's Last Exam (HLE). That test is a collection of extremely difficult expert-level questions across many fields, designed to challenge top AI systems. There are three separate score thresholds here: 35%, 40%, and 45%. A Yes on 45% means the model gets nearly half of these brutally hard questions right. A No means it falls short of that bar. When a new Claude Opus model first appears on the official HLE leaderboard at agi.safe.ai, the market checks its displayed accuracy score the next day at noon Eastern time. If the score is at or above the threshold (35%, 40%, or 45% depending on which market you are looking at), it resolves Yes. If the model disappears from the site before that noon check, that appearance does not count. Only models with 'Opus' in their name qualify. If no Opus model appears on the site by December 31, 2026, the market resolves No. No relevant news was provided for this market. The one headline listed appears to be a poem title and has nothing to do with AI benchmarks. The kind of news that would matter here: an Anthropic announcement of a new Opus model, or an update to the HLE leaderboard showing a new Claude result. The three thresholds tell an interesting story on their own: the market is quite confident a new Opus model will clear 35% (94%), fairly confident about 40% (78%), but genuinely split on 45% (44%). The core unknowns are how much Anthropic improves over current models, when exactly the next Opus is released and tested, and how HLE scores translate across different model configurations. AI capability jumps can be hard to forecast, and the exact number displayed on the site, not any other score, is what decides everything.

The odds right now

• 35%+95%
• 40%+77%
• 45%+44%
• 50%+39%
• 55%+21%

Price history

How this resolves

Resolves December 31, 2026

This market will resolve to "Yes" if the next Claude Opus model added to the Humanity’s Last Exam results at https://agi.safe.ai/ has an HLE Accuracy of at least the specified percentage at 12:00 PM ET on the calendar date following the date on which it first appears on the site. Otherwise, this market will resolve to "No". Read the full resolution rules on the live market page.

Related

Same markets. A fraction of the fee.

These apps all route to the same exchange order book. The difference is what each one adds on top of the exchange's own fee.

Published rates, checked July 2026. MetaMask charges a flat 4 percent per prediction trade. Jupiter adds a fee equal to the exchange's own taker fee at fill time, roughly 2 to 4 percent at typical odds. Where a market carries an exchange settlement fee, it applies everywhere, whichever app you use. Paridesk adds nothing on maker orders.

Trade this market on Paridesk: non-custodial, 0.5% fee.