Granite 4.1 3B vs Llama 3.1 8B Instruct
Wins 1 of 6 areas
Following instructions
Wins 2 of 6 areas
Facts · Long documents
Llama 3.1 8B Instruct wins more areas, narrowly.Granite 4.1 3B is better at following instructions.
Scores updated · 25 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.
- FactsGetting facts right instead of making them up01Llama 3.1 8B Instruct1 of 2 tests · 1 tie
- Long documentsFinding answers in very long texts01Llama 3.1 8B Instruct1 of 1 test
- Following instructionsDoing exactly what it is asked10Granite 4.1 3B1 of 1 test
- CodingWriting and fixing software11Even1 each
- AgentsCarrying out multi-step tasks on its own00Even0 each · 2 ties
- ReasoningHard problems that need careful thinking11Even1 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 three tests each model wins by the widest margin. Scores are out of 100.
Where Llama 3.1 8B Instruct pulls ahead
- Avoids making up answers it doesn't knowAA-Omniscience · Non-hallucination+52.1points ahead
- Reasons across sets of long documentsAA-LCR+12points ahead
- Very hard expert questions across many subjectsHumanity's Last Exam+1.9points ahead
Where Granite 4.1 3B pulls ahead
- Graduate-level biology, physics and chemistry questionsGPQA Diamond+5.5points ahead
- Follows unfamiliar, precisely checkable instructionsIFBench+5.1points ahead
- Hard command-line tasks in a real terminalTerminal-Bench Hard+1.5points ahead
Every test, side by side
All 25 tests both models report. The winning score is in its model's colour; marks a score checked independently.
CodingEven
- Terminal-Bench HardGranite 4.1 3B by 1.52.30.8+1.5
- SciCodeLlama 3.1 8B Instruct by 1.311.913.2+1.3
ReasoningEven
- GPQA DiamondGranite 4.1 3B by 5.531.425.9+5.5
- Humanity's Last ExamLlama 3.1 8B Instruct by 1.93.45.3+1.9
- CritPttie00tie
FactsLlama 3.1 8B Instruct
- AA-Omniscience · Non-hallucinationLlama 3.1 8B Instruct by 52.14.957+52.1
- AA-Omniscience · Accuracytie9.38.5tie
Long documentsLlama 3.1 8B Instruct
- AA-LCRLlama 3.1 8B Instruct by 12618+12
Following instructionsGranite 4.1 3B
- IFBenchGranite 4.1 3B by 5.133.728.6+5.1
Other results14 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.
- AA-OmniscienceLlama 3.1 8B Instruct by 46.1-77-30.9+46.1
- AlpacaEval-2.0Granite 4.1 3B by 11.438.627.2+11.4
- HumanEvalGranite 4.1 3B by 9.181.772.6+9.1
- GSM8KGranite 4.1 3B by 7.186.979.8+7.1
- MMLULlama 3.1 8B Instruct by 66773+6
- HumanEval+Llama 3.1 8B Instruct by 3.476.880.2+3.4
- τ²-Bench Telecom (AA run)Granite 4.1 3B by 3.219.616.4+3.2
- AttaQLlama 3.1 8B Instruct by 1.581.983.4+1.5
- MMLU-ProGranite 4.1 3B by 1.549.848.3+1.5
- Arena HardGranite 4.1 3B by 1.437.836.4+1.4
- MBPPGranite 4.1 3B by 1.171.270.1+1.1
- MGSMGranite 4.1 3B by 1.17068.9+1.1
- AA IntelligenceLlama 3.1 8B Instruct by 15.96.9+1
- Artificial Analysis Coding Indextie4.75.4tie
Questions people ask
Which is better, Granite 4.1 3B or Llama 3.1 8B Instruct?
Llama 3.1 8B Instruct wins two of the six areas where both have results: facts and long documents. Granite 4.1 3B wins following instructions. They are level on coding, agents and reasoning.
Which is better for coding?
Neither. They win 1 coding test each of the 2 both models report.
How do you compare the two?
We use the 25 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 14 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.