The LLM benchmark aggregator.

Granite 4.2 30B

IBM · 30B

Strongest in Knowledge work (#146 of 208), weakest in Agents (#230 of 298). Above par in 2 of 15 scopes.

§ 1 · Profile

What it is good at

Bars run from 50 — the average of the models each source lists — so right of the line is above par. The middle column is the gap to whoever leads that scope.

Human preference50.9−17.9#188/3541/1

Agents44.4−21.9#230/2982/6

Knowledge work41−29.7#146/2082/2

Tool use45.4−28.6—/81✱0/1

Core abilities43.5−22.1—/201✱0/4

Long context26−38.6—/0✱✱

Knowledge49.3−13.2—/197✱0/3

Factuality46.7−14.1—/0✱✱

Reasoning46.9−18.5—/195✱0/5

Science44.7−18.8—/122✱0/1

Expert reasoning30−36.4—/0✱✱

Coding49.1−17—/179✱0/6

Agentic coding48−18.7—/173✱0/5

Code generation50−15.3—/20✱0/1

§ 3 · Sources

Where the numbers come from

4 publications, 12 figures. Every one links to the page it was read from.

LMArena Text ↗1 measureread 2026-11-12

LMArena Text 1340

GDPval-AA ↗1 measureread 2026-11-12

GDPval-AA 583

AA-Briefcase ↗1 measureread 2026-11-12

AA-Briefcase 395

✱✱ IFM — K2 Horizon launch ↗the model’s own publisher9 measuresread 2026-11-03

tau3-Banking 14.4%Terminal-Bench 2.1 26.6%SciCode 36.6%Humanity's Last Exam (without tools) 11.2%GPQA Diamond 64.4%CritPt 0.3%AA-LCR 46.7%AA-Omniscience Accuracy 10.1%AA-Omniscience Non-Hallucination 74.4%

Badge

[![PublicAI Index](https://publicai.io/model-index/badge?model=granite-4-2-30b)](https://publicai.io/model-index/m/granite-4-2-30b)