Anthropic’s November 9 release of Claude Fable 5.2 has driven the current leaderboard, posting 74.3% on Artificial Analysis’s independent HLE run and 72.4% on Scale’s Diamond variant, well ahead of Kimi K3.1 Pro at 70% on tool-augmented evaluations and GPT-6 Astra near 60%. These gains reflect continued scaling of large language models, improved reasoning chains, and selective tool integration on the 2,500-question expert benchmark. Different protocols—closed-book verified sets versus agentic or Diamond subsets—produce score spreads of 10–20 points, creating resolution uncertainty around what counts as the official highest mark by year-end. Remaining catalysts include potential late-2026 launches from OpenAI, Google, or Meta that could close the gap before December 31.
Experimentelle KI-generierte Zusammenfassung mit Polymarket-Daten. Dies ist keine Handelsberatung und spielt keine Rolle bei der Auflösung dieses Marktes. · Aktualisiert1H6H1D7D30DALL
View resolved

Vorsicht bei externen Links.
Vorsicht bei externen Links.
Häufig gestellte Fragen