Llama 3.1 8B Instruct vs Llama 3.1 Instruct 405B
Wins 0 of 6 areas
—
Wins 3 of 6 areas
Coding · Long documents · Following instructions
Llama 3.1 Instruct 405B wins more areas, narrowly.
Scores updated · 43 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 software02Llama 3.1 Instruct 405B2 of 2 tests
- Long documentsFinding answers in very long texts01Llama 3.1 Instruct 405B1 of 1 test
- Following instructionsDoing exactly what it is asked01Llama 3.1 Instruct 405B1 of 1 test
- AgentsCarrying out multi-step tasks on its own00Even0 each · 1 tie
- ReasoningHard problems that need careful thinking11Even1 each · 1 tie
- FactsGetting facts right instead of making them up11Even1 each
Agents, 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+25.6points ahead
- Code for real scientific research problemsSciCode+16.7points ahead
- Answers hard knowledge questions correctlyAA-Omniscience · Accuracy+14.7points ahead
Where Llama 3.1 8B Instruct pulls ahead
- Avoids making up answers it doesn't knowAA-Omniscience · Non-hallucination+9.4points ahead
- Very hard expert questions across many subjectsHumanity's Last Exam+1.3points ahead
Every test, side by side
All 43 tests both models report. The winning score is in its model's colour; marks a score checked independently.
CodingLlama 3.1 Instruct 405B
- SciCodeLlama 3.1 Instruct 405B by 16.713.229.9+16.7
- Terminal-Bench HardLlama 3.1 Instruct 405B by 60.86.8+6
ReasoningEven
- GPQA DiamondLlama 3.1 Instruct 405B by 25.625.951.5+25.6
- Humanity's Last ExamLlama 3.1 8B Instruct by 1.35.34+1.3
- CritPttie00tie
FactsEven
- AA-Omniscience · AccuracyLlama 3.1 Instruct 405B by 14.78.523.2+14.7
- AA-Omniscience · Non-hallucinationLlama 3.1 8B Instruct by 9.45747.6+9.4
Long documentsLlama 3.1 Instruct 405B
- AA-LCRLlama 3.1 Instruct 405B by 7.31825.3+7.3
Following instructionsLlama 3.1 Instruct 405B
- IFBenchLlama 3.1 Instruct 405B by 10.428.639+10.4
Other results33 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 HardLlama 3.1 Instruct 405B by 32.936.469.3+32.9
- DROPLlama 3.1 Instruct 405B by 25.359.584.8+25.3
- MMLU Pro (CoT) (Pass@1 with chain-of-thought)Llama 3.1 Instruct 405B by 2548.373.3+25
- MMLU-ProLlama 3.1 Instruct 405B by 2548.373.3+25
- naturalquestions_closedbookLlama 3.1 Instruct 405B by 24.720.945.6+24.7
- MultiPL-E HumanEvalLlama 3.1 Instruct 405B by 24.450.875.2+24.4
- MGSMLlama 3.1 Instruct 405B by 22.768.991.6+22.7
- MATHLlama 3.1 Instruct 405B by 21.951.973.8+21.9
- MATH (CoT) (Pass@1 with chain-of-thought)Llama 3.1 Instruct 405B by 21.951.973.8+21.9
- OpenBookQALlama 3.1 Instruct 405B by 207494+20
- GSM8KLlama 3.1 Instruct 405B by 1779.896.8+17
- HumanEvalLlama 3.1 Instruct 405B by 16.472.689+16.4
- bbqLlama 3.1 Instruct 405B by 1678.594.5+16
- MBPP EvalPlus (base)Llama 3.1 Instruct 405B by 15.872.888.6+15.8
- BFCL v2Llama 3.1 Instruct 405B by 15.765.481.1+15.7
- MMLULlama 3.1 Instruct 405B by 15.67388.6+15.6
- MMLU (CoT) (Pass@1 with chain-of-thought)Llama 3.1 Instruct 405B by 15.67388.6+15.6
- AA-OmniscienceLlama 3.1 Instruct 405B by 13.8-30.9-17.1+13.8
- ARC-ChallengeLlama 3.1 Instruct 405B by 13.583.496.9+13.5
- MultiPL-E MBPPLlama 3.1 Instruct 405B by 13.352.465.7+13.3
- BFCLv4Llama 3.1 Instruct 405B by 12.476.188.5+12.4
- AlpacaEval-2.0Llama 3.1 Instruct 405B by 12.127.239.3+12.1
- Artificial Analysis Coding IndexLlama 3.1 Instruct 405B by 9.15.414.5+9.1
- IFEvalLlama 3.1 Instruct 405B by 8.280.488.6+8.2
- AIR-Bench 2024Llama 3.1 8B Instruct by 3.762.358.6+3.7
- τ²-Bench Telecom (AA run)Llama 3.1 Instruct 405B by 2.616.419+2.6
- MBPPLlama 3.1 8B Instruct by 1.770.168.4+1.7
- HarmBenchLlama 3.1 Instruct 405B by 1.161.662.7+1.1
- NarrativeQAtie75.674.9tie
- XSTesttie95.395.9tie
- AA Intelligencetie6.97.3tie
- anthropic_red_teamtie96.896.5tie
- simple_safety_teststie98.898.8tie
Questions people ask
Which is better, Llama 3.1 8B Instruct or Llama 3.1 Instruct 405B?
Llama 3.1 Instruct 405B wins three of the six areas where both have results: coding, long documents and following instructions. Llama 3.1 8B Instruct wins none. They are level on agents, reasoning and facts.
Which is better for coding?
Llama 3.1 Instruct 405B. It wins 2 of the 2 coding tests both models report; Llama 3.1 8B Instruct wins none.
How do you compare the two?
We use the 43 benchmark tests both models have published scores on. The verdict counts the 10 tests in the eight capability areas, and a gap under one point (ten on rating-style scales) counts as a tie. The other 33 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.