Llama 3.1 Instruct 405B vs Llama 3.2 3B Instruct
Wins 4 of 5 areas
Coding · Facts · Long documents · Following instructions
Wins 0 of 5 areas
—
Llama 3.1 Instruct 405B is the stronger all-rounder.
Scores updated · 24 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.
- FactsGetting facts right instead of making them up20Llama 3.1 Instruct 405B2 of 2 tests
- CodingWriting and fixing software10Llama 3.1 Instruct 405B1 of 1 test
- Long documentsFinding answers in very long texts10Llama 3.1 Instruct 405B1 of 1 test
- Following instructionsDoing exactly what it is asked10Llama 3.1 Instruct 405B1 of 1 test
- ReasoningHard problems that need careful thinking11Even1 each
Coding, 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 Llama 3.1 Instruct 405B pulls ahead
- Graduate-level biology, physics and chemistry questionsGPQA Diamond+26points ahead
- Code for real scientific research problemsSciCode+24.7points ahead
- Reasons across sets of long documentsAA-LCR+21points ahead
Where Llama 3.2 3B Instruct pulls ahead
- Very hard expert questions across many subjectsHumanity's Last Exam+1.4points ahead
Every test, side by side
All 24 tests both models report. The winning score is in its model's colour; marks a score checked independently.
ReasoningEven
- GPQA DiamondLlama 3.1 Instruct 405B by 2651.525.5+26
- Humanity's Last ExamLlama 3.2 3B Instruct by 1.445.3+1.4
FactsLlama 3.1 Instruct 405B
- AA-Omniscience · AccuracyLlama 3.1 Instruct 405B by 16.923.26.3+16.9
- AA-Omniscience · Non-hallucinationLlama 3.1 Instruct 405B by 5.147.642.5+5.1
Long documentsLlama 3.1 Instruct 405B
- AA-LCRLlama 3.1 Instruct 405B by 2125.34.3+21
Following instructionsLlama 3.1 Instruct 405B
- IFBenchLlama 3.1 Instruct 405B by 12.83926.2+12.8
Other results17 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.
- MATHLlama 3.1 Instruct 405B by 65.373.88.5+65.3
- HumanEvalLlama 3.1 Instruct 405B by 618928+61
- MMLU-ProLlama 3.1 Instruct 405B by 38.373.335+38.3
- BBHLlama 3.1 Instruct 405B by 36.182.946.8+36.1
- MGSMLlama 3.1 Instruct 405B by 33.491.658.2+33.4
- MATH-500 (EM)Llama 3.1 Instruct 405B by 32.673.841.2+32.6
- AA-OmniscienceLlama 3.1 Instruct 405B by 30.4-17.1-47.5+30.4
- MMLULlama 3.1 Instruct 405B by 28.288.660.4+28.2
- MMMLULlama 3.1 Instruct 405B by 25.973.847.9+25.9
- MBPPLlama 3.1 Instruct 405B by 19.768.448.7+19.7
- HellaSwagLlama 3.1 Instruct 405B by 19.489.269.8+19.4
- GSM8KLlama 3.1 Instruct 405B by 19.196.877.7+19.1
- ARC-ChallengeLlama 3.1 Instruct 405B by 18.396.978.6+18.3
- BFCL v2Llama 3.1 Instruct 405B by 14.181.167+14.1
- IFEvalLlama 3.1 Instruct 405B by 11.288.677.4+11.2
- τ²-Bench Telecom (AA run)Llama 3.2 3B Instruct by 2.11921.1+2.1
- AA IntelligenceLlama 3.1 Instruct 405B by 1.67.35.7+1.6
Questions people ask
Which is better, Llama 3.1 Instruct 405B or Llama 3.2 3B Instruct?
Llama 3.1 Instruct 405B wins four of the five areas where both have results: coding, facts, long documents and following instructions. Llama 3.2 3B Instruct wins none. They are level on reasoning.
Which is better for coding?
Llama 3.1 Instruct 405B. It wins the one coding test both models report.
How do you compare the two?
We use the 24 benchmark tests both models have published scores on. The verdict counts the 7 tests in the eight capability areas, and a gap under one point (ten on rating-style scales) counts as a tie. The other 17 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.