**Claude variants currently lead the Humanity’s Last Exam (HLE) leaderboard, with Claude Fable 5 (adaptive reasoning, max effort) at 53.3% and related models like Mythos 5 or Opus 4.8 variants reaching 45–64.5% under optimized settings.** This frontier benchmark comprises 2,500 expert-vetted, graduate-level questions across math, sciences, and humanities, created by the Center for AI Safety and Scale AI to resist saturation. Recent gains stem from Anthropic’s iterative releases and techniques such as extended thinking, tool use (web search, code execution), and fallback mechanisms, which have outpaced GPT-5.4 and Gemini 3.1 Pro previews. With the June 30 resolution date only weeks away, traders are watching for any final Claude updates or configuration tweaks that could push scores higher before cutoff. Historical patterns show Anthropic frequently refines reasoning modes quickly, though major capability jumps often require new model versions rather than incremental tuning.
Experimental AI-generated summary referencing Polymarket data. This is not trading advice and plays no role in how this market resolves. · Updated$361,868 Vol.
45%+
48%
50%+
29%
55%+
8%
$361,868 Vol.
45%+
48%
50%+
29%
55%+
8%
The resolution source will be the official Humanity’s Last Exam leaderboard https://scale.com/leaderboard/humanitys_last_exam.
Market Opened: Jan 30, 2026, 12:00 AM ET
Resolver
0x65070BE91...The resolution source will be the official Humanity’s Last Exam leaderboard https://scale.com/leaderboard/humanitys_last_exam.
Resolver
0x65070BE91...**Claude variants currently lead the Humanity’s Last Exam (HLE) leaderboard, with Claude Fable 5 (adaptive reasoning, max effort) at 53.3% and related models like Mythos 5 or Opus 4.8 variants reaching 45–64.5% under optimized settings.** This frontier benchmark comprises 2,500 expert-vetted, graduate-level questions across math, sciences, and humanities, created by the Center for AI Safety and Scale AI to resist saturation. Recent gains stem from Anthropic’s iterative releases and techniques such as extended thinking, tool use (web search, code execution), and fallback mechanisms, which have outpaced GPT-5.4 and Gemini 3.1 Pro previews. With the June 30 resolution date only weeks away, traders are watching for any final Claude updates or configuration tweaks that could push scores higher before cutoff. Historical patterns show Anthropic frequently refines reasoning modes quickly, though major capability jumps often require new model versions rather than incremental tuning.
Experimental AI-generated summary referencing Polymarket data. This is not trading advice and plays no role in how this market resolves. · Updated


Beware of external links.
Beware of external links.
Frequently Asked Questions