Home Model series Gemini Flash Gemini Flash · Humanity's Last Exam · HLE-Verified 1,811-item verified set · Release history Gemini Flash · Humanity's Last Exam · HLE-Verified 1,811-item verified set · Release history View scores, configurations, and disclosures for Humanity's Last Exam · HLE-Verified 1,811-item verified set across 2 Gemini Flash releases, ordered by release date.
Differences are calculated only for unique, active, exact values between direct predecessors when the unit, metric, and explicitly recorded configuration and conditions match.
2026-08-13 · 1 score record for this release, including 1 exact value (may include historical records)
Gemini 3.7 Flash · Humanity's Last Exam scores and configurations Metric and score Configuration and evaluation conditions Sources Accuracy 53.6% Reported 53.6% · Unit % Published Unconfirmed
Gemini 3.7 Flash default sampling Settings Gemini API gemini-3.7-flash; default sampling; pass@1 except where noted. Unknowns Exact sampling parameters are not published. Tools Tool access not recorded Reasoning budget not recorded Framework not recorded Protocol Self-computed on HLE-Verified 1,811-item set; excludes 689 uncertain original items. Conditions Pass@1; results as of August 2026. View this score in rankings
PDF p.4, HLE-Verified row; method p.3 Attachment · Gemini 3.7 Flash evaluation methodology and results 2026-09-02 · 1 score record for this release, including 1 exact value (may include historical records)
Records for the direct predecessor do not provide matching conditions for a valid difference.
Gemini 3.8 Flash · Humanity's Last Exam scores and configurations Metric and score Configuration and evaluation conditions Sources Accuracy 54.9% Reported 54.9% · Unit % Published Unconfirmed
Gemini 3.8 Flash default sampling Settings Gemini API gemini-3.8-flash; default sampling; pass@1 except where noted. Unknowns Exact sampling parameters are not published. Tools Tool access not recorded Reasoning budget not recorded Framework not recorded Protocol Self-computed accuracy on 1,811 verified items: 668 verified original HLE items plus 1,143 revised items; 689 uncertain original items excluded. Conditions Gemini API, default sampling, pass@1; results as of September 2026. View this score in rankings
PDF p.4, HLE-Verified row; method p.3 Attachment · Gemini 3.8 Flash evaluation methodology and results