Granite 4.1 3B — Real Scores Across 8 Benchmarks (2026)

IBM

Granite 4.1 3B

8 benchmarks

GSM8K

#6 of 6

86.88%

HumanEval

#13 of 22

81.71%

HumanEval+

#5 of 5

76.83%

MMLU

#3 of 3

67.02%

BFCL v3

#5 of 5

60.8%

MMMLU

#12 of 12

57.61%

MMLU Pro

#55 of 58

49.83%

GPQA Diamond

#117 of 124

31.7%