Granite 4.1 30B vs Granite 4.1 3B
Wins 6 of 6 areas
Coding · Agents · Reasoning · Facts · Long documents · Following instructions
Wins 0 of 6 areas
—
Granite 4.1 30B is the stronger all-rounder.
Scores updated · 44 tests both models report · How we compare
Where each one wins
Tests won in each of the six areas where both have results. Each piece is one test, so a longer bar means more evidence; grey means the two scored within a point of each other.
- ReasoningHard problems that need careful thinking10Granite 4.1 30B1 of 3 tests · 2 ties
- CodingWriting and fixing software10Granite 4.1 30B1 of 2 tests · 1 tie
- AgentsCarrying out multi-step tasks on its own10Granite 4.1 30B1 of 2 tests · 1 tie
- FactsGetting facts right instead of making them up10Granite 4.1 30B1 of 2 tests · 1 tie
- Long documentsFinding answers in very long texts10Granite 4.1 30B1 of 1 test
- Following instructionsDoing exactly what it is asked10Granite 4.1 30B1 of 1 test
Long documents and following instructions rest on a single test each.
The biggest differences
The tests each model wins by the widest margin, up to three each. Scores are out of 100.
Where Granite 4.1 30B pulls ahead
- Reasons across sets of long documentsAA-LCR+17.7points ahead
- Graduate-level biology, physics and chemistry questionsGPQA Diamond+16.7points ahead
- Code for real scientific research problemsSciCode+13.9points ahead
Where Granite 4.1 3B pulls ahead
No clear win on a test scored out of 100.
Every test, side by side
All 44 tests both models report. The winning score is in its model's colour; marks a score checked independently.
CodingGranite 4.1 30B
- SciCodeGranite 4.1 30B by 13.925.811.9+13.9
- Terminal-Bench Hardtie2.32.3tie
AgentsGranite 4.1 30B
- τ-Bench V3 · BankingGranite 4.1 30B by 2.73.91.2+2.7
- GDPValtie00tie
ReasoningGranite 4.1 30B
- GPQA DiamondGranite 4.1 30B by 16.748.131.4+16.7
- Humanity's Last Examtie4.13.4tie
- CritPttie00tie
FactsGranite 4.1 30B
- AA-Omniscience · AccuracyGranite 4.1 30B by 4.713.99.3+4.7
- AA-Omniscience · Non-hallucinationtie54.9tie
Following instructionsGranite 4.1 30B
- IFBenchGranite 4.1 30B by 10.744.433.7+10.7
Other results33 tests, not counted
Tests outside the eight areas. They are not counted above: several are summary scores built from other tests, or the same test under another name.
- Arena HardGranite 4.1 30B by 33.27137.8+33.2
- τ²-Bench Telecom (AA run)Granite 4.1 30B by 22.542.119.6+22.5
- AlpacaEval-2.0Granite 4.1 30B by 17.656.238.6+17.6
- DeepMind MathGranite 4.1 30B by 17.381.964.6+17.3
- MMMLUGranite 4.1 30B by 16.173.757.6+16.1
- Eval+ AvgGranite 4.1 30B by 15.682.767+15.6
- IncludeGranite 4.1 30B by 15.267.352+15.2
- CRUXEval-O (output prediction)Granite 4.1 30B by 1555.840.8+15
- MBPPGranite 4.1 30B by 14.385.571.2+14.3
- MMLU-ProGranite 4.1 30B by 14.364.149.8+14.3
- GPQA (unspecified)Granite 4.1 30B by 14.145.831.7+14.1
- Minerva MathGranite 4.1 30B by 13.481.367.9+13.4
- MMLUGranite 4.1 30B by 13.280.267+13.2
- BFCL v3Granite 4.1 30B by 12.973.760.8+12.9
- AGI EVALGranite 4.1 30B by 12.677.865.2+12.6
- Tulu3 Safety Eval AvgGranite 4.1 30B by 11.478.266.8+11.4
- MBPP+ (EvalPlus-augmented)Granite 4.1 30B by 11.373.562.2+11.3
- MultiPL-EGranite 4.1 30B by 9.862.352.5+9.8
- AA-OmniscienceGranite 4.1 30B by 9.2-67.8-77+9.2
- HumanEval+Granite 4.1 30B by 8.685.476.8+8.6
- BBHGranite 4.1 30B by 7.983.775.8+7.9
- IFEval (avg)Granite 4.1 30B by 7.489.782.3+7.4
- GSM8KGranite 4.1 30B by 7.394.286.9+7.3
- HumanEvalGranite 4.1 30B by 6.788.481.7+6.7
- BigCodeBenchGranite 4.1 30B by 6.638.832.2+6.6
- Artificial Analysis Coding IndexGranite 4.1 30B by 5.710.44.7+5.7
- GSM SymbolicGranite 4.1 3B by 5.675.781.3+5.6
- AttaQGranite 4.1 30B by 3.985.881.9+3.9
- SimpleQAGranite 4.1 30B by 3.16.83.7+3.1
- SALAD-BenchGranite 4.1 30B by 2.496.494+2.4
- AA IntelligenceGranite 4.1 30B by 1.57.45.9+1.5
- MGSMGranite 4.1 30B by 1.171.170+1.1
- MTBench AvgGranite 4.1 30B by 18.67.6+1
Questions people ask
Which is better, Granite 4.1 30B or Granite 4.1 3B?
Granite 4.1 30B wins all six areas where both have results: coding, agents, reasoning, facts, long documents and following instructions. Granite 4.1 3B wins none.
Which is better for coding?
Granite 4.1 30B. It wins 1 of the 2 coding tests both models report; Granite 4.1 3B wins none, and 1 is a tie.
How do you compare the two?
We use the 44 benchmark tests both models have published scores on. The verdict counts the 11 tests in the eight capability areas, and a gap under one point (ten on rating-style scales) counts as a tie. The other 33 are listed but not counted, because several are summary scores or repeat a test. Each score is the one shown on the model's own page.