Google DeepMind's Gemini 3.1 Pro Preview currently leads the Humanity's Last Exam leaderboard at 45.9% accuracy, a frontier benchmark of 2,500 expert-level questions testing advanced reasoning across math, science, and humanities, outpacing OpenAI's GPT-5 Pro (31.6%) and Anthropic's Claude Opus 4.6 (34.4%). This positioning stems from March 2026 releases like Gemini 3 Deep Think, which hit 41% without tools via parallel reasoning chains, doubling prior scores amid intensifying AI lab competition. Today's Gemma 4 open models, built on Gemini 3 architecture, signal further reasoning gains. Traders eye Google I/O in May for potential Gemini 4 previews that could breach 50% by June 30, though calibration errors and rapid benchmark evolution introduce uncertainty.
Resumo experimental gerado por IA com dados do Polymarket · AtualizadoPontuação do Google Gemini no último exame da Humanidade até 30 de junho?
Pontuação do Google Gemini no último exame da Humanidade até 30 de junho?
$264,399 Vol.
40%+
93%
45%+
72%
50%+
39%
55%+
16%
60%+
10%
$264,399 Vol.
40%+
93%
45%+
72%
50%+
39%
55%+
16%
60%+
10%
The resolution source will be the official Humanity’s Last Exam leaderboard https://scale.com/leaderboard/humanitys_last_exam.
Mercado Aberto: Jan 29, 2026, 12:50 PM ET
Resolver
0x65070BE91...The resolution source will be the official Humanity’s Last Exam leaderboard https://scale.com/leaderboard/humanitys_last_exam.
Resolver
0x65070BE91...Google DeepMind's Gemini 3.1 Pro Preview currently leads the Humanity's Last Exam leaderboard at 45.9% accuracy, a frontier benchmark of 2,500 expert-level questions testing advanced reasoning across math, science, and humanities, outpacing OpenAI's GPT-5 Pro (31.6%) and Anthropic's Claude Opus 4.6 (34.4%). This positioning stems from March 2026 releases like Gemini 3 Deep Think, which hit 41% without tools via parallel reasoning chains, doubling prior scores amid intensifying AI lab competition. Today's Gemma 4 open models, built on Gemini 3 architecture, signal further reasoning gains. Traders eye Google I/O in May for potential Gemini 4 previews that could breach 50% by June 30, though calibration errors and rapid benchmark evolution introduce uncertainty.
Resumo experimental gerado por IA com dados do Polymarket · Atualizado
Cuidado com os links externos.
Cuidado com os links externos.
Frequently Asked Questions