Granite 4.1 3B vs Qwen2.5 Instruct 72B
Wins 0 of 5 areas
—
Wins 5 of 5 areas
Coding · Reasoning · Facts · Long documents · Following instructions
Qwen2.5 Instruct 72B is the stronger all-rounder.
Scores updated · 27 tests both models report · How we compare
Where each one wins
Tests won in each of the five 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 software02Qwen2.5 Instruct 72B2 of 2 tests
- FactsGetting facts right instead of making them up02Qwen2.5 Instruct 72B2 of 2 tests
- ReasoningHard problems that need careful thinking01Qwen2.5 Instruct 72B1 of 3 tests · 2 ties
- Long documentsFinding answers in very long texts01Qwen2.5 Instruct 72B1 of 1 test
- Following instructionsDoing exactly what it is asked01Qwen2.5 Instruct 72B1 of 1 test
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 Qwen2.5 Instruct 72B pulls ahead
- Graduate-level biology, physics and chemistry questionsGPQA Diamond+17.7points ahead
- Code for real scientific research problemsSciCode+14.8points ahead
- Reasons across sets of long documentsAA-LCR+14points ahead
Where Granite 4.1 3B pulls ahead
No clear win on a test scored out of 100.
Every test, side by side
All 27 tests both models report. The winning score is in its model's colour; marks a score checked independently.
CodingQwen2.5 Instruct 72B
- SciCodeQwen2.5 Instruct 72B by 14.811.926.7+14.8
- Terminal-Bench HardQwen2.5 Instruct 72B by 2.22.34.5+2.2
ReasoningQwen2.5 Instruct 72B
- GPQA DiamondQwen2.5 Instruct 72B by 17.731.449.1+17.7
- Humanity's Last Examtie3.43.6tie
- CritPttie00tie
FactsQwen2.5 Instruct 72B
- AA-Omniscience · Non-hallucinationQwen2.5 Instruct 72B by 9.64.914.5+9.6
- AA-Omniscience · AccuracyQwen2.5 Instruct 72B by 8.29.317.5+8.2
Following instructionsQwen2.5 Instruct 72B
- IFBenchQwen2.5 Instruct 72B by 3.233.736.9+3.2
Other results18 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 HardQwen2.5 Instruct 72B by 43.437.881.2+43.4
- AA-OmniscienceQwen2.5 Instruct 72B by 23.9-77-53.1+23.9
- MultiPL-EQwen2.5 Instruct 72B by 22.652.575.1+22.6
- MMLU-ProQwen2.5 Instruct 72B by 21.349.871.1+21.3
- CRUXEval-O (output prediction)Qwen2.5 Instruct 72B by 19.140.859.9+19.1
- MMMLUQwen2.5 Instruct 72B by 17.257.674.8+17.2
- MBPPQwen2.5 Instruct 72B by 1771.288.2+17
- τ²-Bench Telecom (AA run)Qwen2.5 Instruct 72B by 14.919.634.5+14.9
- MBPP+ (EvalPlus-augmented)Qwen2.5 Instruct 72B by 14.862.277+14.8
- MMLUQwen2.5 Instruct 72B by 106777+10
- GSM8KQwen2.5 Instruct 72B by 8.986.995.8+8.9
- Artificial Analysis Coding IndexQwen2.5 Instruct 72B by 7.24.711.9+7.2
- MGSMQwen2.5 Instruct 72B by 6.27076.2+6.2
- SimpleQAQwen2.5 Instruct 72B by 5.43.79.1+5.4
- HumanEvalQwen2.5 Instruct 72B by 4.981.786.6+4.9
- IFEval (avg)Qwen2.5 Instruct 72B by 4.982.387.2+4.9
- BBHQwen2.5 Instruct 72B by 475.879.8+4
- AA IntelligenceQwen2.5 Instruct 72B by 1.85.97.7+1.8
Questions people ask
Which is better, Granite 4.1 3B or Qwen2.5 Instruct 72B?
Qwen2.5 Instruct 72B wins all five areas where both have results: coding, reasoning, facts, long documents and following instructions. Granite 4.1 3B wins none.
Which is better for coding?
Qwen2.5 Instruct 72B. It wins 2 of the 2 coding tests both models report; Granite 4.1 3B wins none.
How do you compare the two?
We use the 27 benchmark tests both models have published scores on. The verdict counts the 9 tests in the eight capability areas, and a gap under one point (ten on rating-style scales) counts as a tie. The other 18 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.