Granite 4.1 8B vs Qwen2.5 Instruct 72B
Wins 1 of 5 areas
Following instructions
Wins 4 of 5 areas
Coding · Reasoning · Facts · Long documents
Qwen2.5 Instruct 72B is the stronger all-rounder.Granite 4.1 8B is cheaper and better at following instructions.
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 asked10Granite 4.1 8B1 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.
Granite 4.1 8B costs 85% less for the same work.
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
- Reasons across sets of long documentsAA-LCR+6.7points ahead
- Graduate-level biology, physics and chemistry questionsGPQA Diamond+5.8points ahead
- Answers hard knowledge questions correctlyAA-Omniscience · Accuracy+5.2points ahead
Where Granite 4.1 8B pulls ahead
- Follows unfamiliar, precisely checkable instructionsIFBench+1.7points ahead
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 4.921.826.7+4.9
- Terminal-Bench HardQwen2.5 Instruct 72B by 4.504.5+4.5
ReasoningQwen2.5 Instruct 72B
- GPQA DiamondQwen2.5 Instruct 72B by 5.843.349.1+5.8
- Humanity's Last Examtie3.83.6tie
- CritPttie00tie
FactsQwen2.5 Instruct 72B
- AA-Omniscience · AccuracyQwen2.5 Instruct 72B by 5.212.317.5+5.2
- AA-Omniscience · Non-hallucinationQwen2.5 Instruct 72B by 1.612.914.5+1.6
Long documentsQwen2.5 Instruct 72B
- AA-LCRQwen2.5 Instruct 72B by 6.713.320+6.7
Following instructionsGranite 4.1 8B
- IFBenchGranite 4.1 8B by 1.738.636.9+1.7
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.
- MMLU-ProQwen2.5 Instruct 72B by 15.15671.1+15.1
- MultiPL-EQwen2.5 Instruct 72B by 14.860.375.1+14.8
- CRUXEval-O (output prediction)Qwen2.5 Instruct 72B by 12.347.659.9+12.3
- Arena HardQwen2.5 Instruct 72B by 12.26981.2+12.2
- AA-OmniscienceQwen2.5 Instruct 72B by 11-64.1-53.1+11
- MMMLUQwen2.5 Instruct 72B by 1064.874.8+10
- τ²-Bench Telecom (AA run)Qwen2.5 Instruct 72B by 6.727.834.5+6.7
- MGSMGranite 4.1 8B by 6.182.376.2+6.1
- SimpleQAQwen2.5 Instruct 72B by 4.34.89.1+4.3
- GSM8KQwen2.5 Instruct 72B by 3.392.595.8+3.3
- MBPP+ (EvalPlus-augmented)Qwen2.5 Instruct 72B by 3.273.877+3.2
- MMLUQwen2.5 Instruct 72B by 3.273.877+3.2
- Artificial Analysis Coding IndexQwen2.5 Instruct 72B by 2.49.511.9+2.4
- HumanEvalQwen2.5 Instruct 72B by 1.285.486.6+1.2
- AA IntelligenceQwen2.5 Instruct 72B by 1.16.67.7+1.1
- MBPPtie87.388.2tie
- BBHtie80.579.8tie
- IFEval (avg)tie87.187.2tie
Questions people ask
Which is better, Granite 4.1 8B or Qwen2.5 Instruct 72B?
Qwen2.5 Instruct 72B wins four of the five areas where both have results: coding, reasoning, facts and long documents. Granite 4.1 8B wins following instructions, and costs 85% less.
Which is better for coding?
Qwen2.5 Instruct 72B. It wins 2 of the 2 coding tests both models report; Granite 4.1 8B wins none.
Which is cheaper?
Granite 4.1 8B costs $0.05 per million input tokens and $0.10 per million output tokens; Qwen2.5 Instruct 72B costs $0.47 and $0.49. That makes Granite 4.1 8B about 85% cheaper for the same work.
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.