Gemma 3 4B vs Granite 4.1 3B
Wins 0 of 6 areas
—
Wins 3 of 6 areas
Coding · Facts · Following instructions
Granite 4.1 3B wins more areas, narrowly.
Scores updated · 26 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.
- CodingWriting and fixing software02Granite 4.1 3B2 of 2 tests
- FactsGetting facts right instead of making them up02Granite 4.1 3B2 of 2 tests
- Following instructionsDoing exactly what it is asked01Granite 4.1 3B1 of 1 test
- AgentsCarrying out multi-step tasks on its own00Even0 each · 2 ties
- ReasoningHard problems that need careful thinking11Even1 each · 1 tie
- Long documentsFinding answers in very long texts00Even0 each · 1 tie
Long documents and following instructions rest on a single test each.
What it costs
Prices per million tokens, roughly 750,000 words. The bars show the cost of a million tokens read plus a million written.
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 3B pulls ahead
- Follows unfamiliar, precisely checkable instructionsIFBench+5.4points ahead
- Code for real scientific research problemsSciCode+4.6points ahead
- Avoids making up answers it doesn't knowAA-Omniscience · Non-hallucination+3.1points ahead
Where Gemma 3 4B pulls ahead
- Very hard expert questions across many subjectsHumanity's Last Exam+1.9points ahead
Every test, side by side
All 26 tests both models report. The winning score is in its model's colour; marks a score checked independently.
CodingGranite 4.1 3B
- SciCodeGranite 4.1 3B by 4.67.311.9+4.6
- Terminal-Bench HardGranite 4.1 3B by 1.50.82.3+1.5
ReasoningEven
- GPQA DiamondGranite 4.1 3B by 2.329.131.4+2.3
- Humanity's Last ExamGemma 3 4B by 1.95.33.4+1.9
- CritPttie00tie
FactsGranite 4.1 3B
- AA-Omniscience · Non-hallucinationGranite 4.1 3B by 3.11.84.9+3.1
- AA-Omniscience · AccuracyGranite 4.1 3B by 1.67.79.3+1.6
Following instructionsGranite 4.1 3B
- IFBenchGranite 4.1 3B by 5.428.333.7+5.4
Other results15 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.
- MGSMGemma 3 4B by 17.387.370+17.3
- τ²-Bench Telecom (AA run)Granite 4.1 3B by 14.6519.6+14.6
- HumanEval+Granite 4.1 3B by 1462.876.8+14
- HumanEvalGranite 4.1 3B by 10.471.381.7+10.4
- MMLUGranite 4.1 3B by 8.658.467+8.6
- MBPPGranite 4.1 3B by 863.271.2+8
- MMMLUGranite 4.1 3B by 7.550.157.6+7.5
- MMLU-ProGranite 4.1 3B by 6.243.649.8+6.2
- Arena HardGranite 4.1 3B by 631.837.8+6
- AA-OmniscienceGranite 4.1 3B by 5.9-82.9-77+5.9
- BBHGranite 4.1 3B by 3.672.275.8+3.6
- GSM8KGemma 3 4B by 2.389.286.9+2.3
- Artificial Analysis Coding IndexGranite 4.1 3B by 22.74.7+2
- AA IntelligenceGranite 4.1 3B by 1.14.85.9+1.1
- SimpleQAtie43.7tie
Questions people ask
Which is better, Gemma 3 4B or Granite 4.1 3B?
Granite 4.1 3B wins three of the six areas where both have results: coding, facts and following instructions. Gemma 3 4B wins none. They are level on agents, reasoning and long documents.
Which is better for coding?
Granite 4.1 3B. It wins 2 of the 2 coding tests both models report; Gemma 3 4B wins none.
How do you compare the two?
We use the 26 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 15 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.