Inkling — Real Scores Across 14 Benchmarks (2026)

Thinking Machines Lab

Inkling

14 benchmarks

AIME

#10 of 117

97.1%

GPQA Diamond

#43 of 118

87.2%

IFBench

#3 of 3

79.8%

CharXiv

#18 of 23

78.1%

SWE-bench Verified

#21 of 86

77.6%

BrowseComp

#26 of 45

77.1%

MCP-Atlas

#14 of 17

74.1%

MMMU Pro

#25 of 36

73.5%

Terminal-Bench 2.1

#22 of 27

63.8%

SWE-bench Pro

#27 of 43

54.3%

Humanity's Last Exam

#27 of 72

46.0%

Toolathlon-Verified

#7 of 7

45.5%

SimpleQA Verified

#3 of 5

43.9%

Humanity's Last Exam

#27 of 72

29.7%