Benchmark Almanac

granite-4.0-1b

granite-4.0-1b.

Released October 28th, 2025 Compare with others
24
Benchmarks with a score
0
Ranked first, out of five or more models
0
In the top three
Unknown lab Oct 2025

granite-4.0-1b

106.5 #386 of 404
Self-Reported Index, from the lab's own numbers. No Epoch Capabilities Index score yet.
  • GPQA Diamond 24% #222 /229
  • BigCodeBench 30.2% #14 /14
  • MMLU-Pro 34% #164 /167

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 49.2% #27 of 40 Lab model card
BIG-Bench Hard 60.4% #59 of 73 Lab model card
IFEval 77.4% #114 of 135 Lab model card
MMLU 59.4% #118 of 131 Lab model card
MGSM 37.8% #55 of 57 Lab model card
INCLUDE 42.1% #30 of 31 Lab model card
MMMLU 45% #69 of 71 Lab model card
MMLU-Pro 34% #164 of 167 Lab model card

Math and science

BenchmarkScoreRankSource
GSM8K 76.3% #70 of 92 Lab model card
MATH 45.3% #76 of 99 Lab model card
GPQA Diamond 24% #222 of 229 Epoch AI run
OTIS Mock AIME 2024-2025 0.83% #206 of 212 Epoch AI run
DeepMind Mathematics 34% #4 of 4 Lab model card
GSM-Symbolic 72.3% #3 of 3 Lab model card

Coding

BenchmarkScoreRankSource
HumanEval 74% #66 of 101 Lab model card
MBPP 65% #51 of 71 Lab model card
HumanEval+ 69% #24 of 27 Lab model card
MultiPL-E 32.2% #18 of 18 Lab model card
MBPP+ 57% #15 of 15 Lab model card
BigCodeBench 30.2% #14 of 14 Lab model card
CRUXEval 33.1% #9 of 9 Lab model card

Agents and tools

BenchmarkScoreRankSource
Berkeley Function Calling Leaderboard (BFCL) 54.8% #40 of 41 Lab model card

Safety and security

BenchmarkScoreRankSource
AttaQ 85.3% #5 of 5 Lab model card
SALAD-Bench 93.4% #3 of 3 Lab model card

Its lab's numbers put it at 106.5, number 386 of 404

The Self-Reported Index fits one score to the 19 capability results the lab 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 104.2 to 108.8. How we compute it.

Its lab's number next to a leaderboard

Where the benchmark's own leaderboard, or Epoch AI's run, also has this model. Gaps mostly come from different harnesses, effort levels and task subsets. Every pair.

Benchmark Lab says Board says Gap
GPQA Diamond 29.9% GPQA (0-shot, CoT; subset not stated) 24% Epoch AI run +5.9 pts

Its launch post

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

granite-4.0-1b model card (Hugging Face) Oct 28, 2025