Will Google Gemini score at least 55% on Humanity’s Last Exam?

Google DeepMind (Hassabis/Pichai/Gemini team) controls releases; current Gemini 3.1 Pro Preview tops HLE at 46.44% with no announced major update or new model in pipeline for the next 22 days. Historical pattern shows multi-month gaps between frontier releases and incremental gains of 5-10pp, not 9pp jumps in weeks. No incentive or capability exists to force a 55% score by deadline; base rate of such rapid benchmark leaps is near zero.

Rules require Yes only if official Scale leaderboard explicitly lists a Gemini model at >=55% by 11:59 PM ET June 30 2026; current top Gemini entry is 46.44% (Gemini 3.1 Pro Preview, thinking-high variant) per multiple sources including labs.scale.com and artificialanalysis.ai as of June 7. No Gemini release or evaluation update capable of a 9+ point jump is documented or scheduled before the hard deadline, satisfying the strict evidence standard and default-No on ambiguity. Smart-money NO lean at 0.87 is consistent with the literal reading; no resolver update or attribution loophole exists.

Official Scale leaderboard shows Gemini 3.1 Pro Preview at 46.44% (thinking high) as of latest data, with other sources confirming ~44.7-46.4%; no new Gemini releases or jumps evident in last 72h searches. Smart-money whales heavily favor No ($221 stake vs $13 on Yes, 0.87 confidence), consistent with 86¢ No price and $63k volume. 22 days remain to June 30 deadline with gradual prior gains (37%→46% over months) making 55% threshold unreachable without unprecedented leap, supporting consensus No hold.