Benchmark Almanac

Granite 4.1 30B

Granite 4.1 30B places in the top three on AttaQ.

IBM Released April 29th, 2026 Compare with others
28
Benchmarks with a score
0
Ranked first, out of five or more models
1
In the top three
IBM Apr 2026

Granite 4.1 30B

Open weights

124.0 #321 of 404
Self-Reported Index, from the lab's own numbers. No Epoch Capabilities Index score yet.
  • BigCodeBench 38.8% #13 /14
  • MMLU-Pro 64.1% #140 /167
  • IFEval 89.7% #37 /135

Where it ranks

Its best score on each benchmark, and where that ranks among every model with a score there. We rank each model once, at its best setting.

Knowledge and language

BenchmarkScoreRankSource
AGIEval 77.8% #9 of 40 Lab model card
IFEval 89.7% #37 of 135 Lab model card
BIG-Bench Hard 83.7% #28 of 73 Lab model card
AlpacaEval 2.0 56.2% #7 of 16 Lab model card
MMLU 80.2% #76 of 131 Lab model card
Arena-Hard-Auto 71% #26 of 39 Lab model card
MT-Bench 8.61 #16 of 23 Lab model card
INCLUDE 67.3% #23 of 31 Lab model card
MGSM 71.1% #43 of 57 Lab model card
MMMLU 73.7% #56 of 71 Lab model card
MMLU-Pro 64.1% #140 of 167 Lab model card
SimpleQA 6.81% #54 of 60 Lab model card

Math and science

BenchmarkScoreRankSource
GSM8K 94.2% #19 of 92 Lab model card
MATH 81.3% #23 of 99 Lab model card
DeepMind Mathematics 81.9% #1 of 4 Lab model card
GSM-Symbolic 75.7% #2 of 3 Lab model card
GPQA Diamond 45.8% #257 of 302 Lab model card

Coding

BenchmarkScoreRankSource
MBPP 85.5% #10 of 71 Lab model card
HumanEval 88.4% #19 of 101 Lab model card
MBPP+ 73.5% #8 of 15 Lab model card
HumanEval+ 85.4% #15 of 27 Lab model card
CRUXEval 55.8% #6 of 9 Lab model card
MultiPL-E 62.3% #16 of 18 Lab model card
BigCodeBench 38.8% #13 of 14 Lab model card

Agents and tools

BenchmarkScoreRankSource
Berkeley Function Calling Leaderboard (BFCL) 73.7% #8 of 41 Lab model card

Safety and security

BenchmarkScoreRankSource
AttaQ 85.8% #3 of 5 Lab model card
SALAD-Bench 96.4% #2 of 3 Lab model card
Tulu 3 Safety Eval 78.2% #5 of 6 Lab model card

Its lab's numbers put it at 124.0, number 321 of 404

The Self-Reported Index fits one score to the 22 capability results IBM reported for this model, on the scale of the Epoch Capabilities Index. Labs choose which benchmarks to show and how to run them, so read it as the lab's claim, not an independent measurement. The 90% interval runs from 121.9 to 126.0. How we compute it.

Its launch post

The benchmarks the lab chose to headline, with the scores it gave.

granite-4.1-30b model card (Hugging Face) Apr 29, 2026