Frontier Log

GoogleModel

Gemini 3 Deep Think

Made by
Google
Released
Announced in
Gemini 3 Deep Think: Advancing science, research and engineeringblog.google
Event
Gemini 3 Deep Think upgrade
API record
/api/v1/models/gemini-3-deep-think

Benchmark results

BenchmarkScoreSource check
Humanity's Last ExamNo tools48.4%
CheckedA person checked this number against its source on 20 Oct 2026. This confirms the number matches what the lab published, not that the score was reproduced.
Humanity's Last ExamWith tools · search and code execution53.4%
CheckedA person checked this number against its source on 20 Oct 2026. This confirms the number matches what the lab published, not that the score was reproduced.
ARC-AGI-2ARC Prize Verified84.6%
CheckedA person checked this number against its source on 20 Oct 2026. This confirms the number matches what the lab published, not that the score was reproduced.
Sources

Reported by Google

Events

All models