Gemini 3.6 Flash — Real Scores Across 9 Benchmarks (2026)

Google DeepMind

Gemini 3.6 Flash

9 benchmarks

GDM-MRCR v2

#1 of 7

91.8%

CharXiv With Tools

#1 of 4

89.4%

CharXiv

#7 of 32

85.2%

OSWorld-Verified

#2 of 19

83.0%

Terminal-Bench 2.1

#19 of 34

78.0%

MLE-Bench

#2 of 6

63.9%

SWE-bench Pro

#17 of 49

58.7%

GDM-MRCR v2 1M

#1 of 3

54.0%

DeepSWE

#16 of 23

49.0%