Granite 4.1 30B

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

IBM Released 29 Apr 2026

Benchmarks with a score

Ranked first, out of five or more models

In the top three

IBM Apr 2026

Open weights

No Epoch Capabilities Index score yet

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.

Benchmark	Score	Rank	Source
AGIEval	77.8%	#7 of 35	Model card
IFEval	89.7%	#30 of 126	Model card
BIG-Bench Hard	83.7%	#26 of 71	Model card
AlpacaEval 2.0	56.2%	#7 of 16	Model card
MMLU	80.2%	#74 of 127	Model card
Arena-Hard-Auto	71%	#26 of 39	Model card
MT-Bench	8.61	#16 of 23	Model card
MGSM	71.1%	#38 of 52	Model card
INCLUDE	67.3%	#19 of 25	Model card
MMMLU	73.7%	#50 of 62	Model card
MMLU-Pro	64.1%	#126 of 151	Model card
SimpleQA	6.81%	#53 of 59	Model card

Benchmark	Score	Rank	Source
MATH	81.3%	#18 of 93	Model card
GSM8K	94.2%	#19 of 87	Model card
DeepMind Mathematics	81.9%	#1 of 4	Model card
GSM-Symbolic	75.7%	#2 of 3	Model card
GPQA Diamond	45.8%	#242 of 284	Model card
CritPt	0%	#157 of 160	Epoch AI

Benchmark	Score	Rank	Source
MBPP	85.5%	#8 of 66	Model card
HumanEval	88.4%	#19 of 95	Model card
MBPP+	73.5%	#8 of 15	Model card
HumanEval+	85.4%	#15 of 27	Model card
CRUXEval	55.8%	#6 of 9	Model card
MultiPL-E	62.3%	#16 of 18	Model card
BigCodeBench	38.8%	#10 of 11	Model card

Benchmark	Score		Rank	Source
Berkeley Function Calling Leaderboard (BFCL)	73.7%		#13 of 47	Model card

Benchmark	Score	Rank	Source
AttaQ	85.8%	#3 of 5	Model card
SALAD-Bench	96.4%	#2 of 3	Model card
Tulu 3 Safety Eval	78.2%	#5 of 6	Model card

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