Claude 3 Haiku vs Llama 3.1 Instruct 405B
Wins 1 of 6 areas
Long documents
Wins 4 of 6 areas
Coding · Reasoning · Facts · Following instructions
Llama 3.1 Instruct 405B is the stronger all-rounder.Claude 3 Haiku is better at long documents.
Scores updated · 32 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
- FactsGetting facts right instead of making them up02Llama 3.1 Instruct 405B2 of 2 tests
- ReasoningHard problems that need careful thinking01Llama 3.1 Instruct 405B1 of 3 tests · 2 ties
- Following instructionsDoing exactly what it is asked01Llama 3.1 Instruct 405B1 of 1 test
- Long documentsFinding answers in very long texts10Claude 3 Haiku1 of 1 test
- AgentsCarrying out multi-step tasks on its own00Even0 each · 1 tie
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
- Avoids making up answers it doesn't knowAA-Omniscience · Non-hallucination+28.1points ahead
- Graduate-level biology, physics and chemistry questionsGPQA Diamond+14.1points ahead
- Code for real scientific research problemsSciCode+11.3points ahead
Where Claude 3 Haiku pulls ahead
- Reasons across sets of long documentsAA-LCR+2.4points ahead
Every test, side by side
All 32 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 11.318.629.9+11.3
- Terminal-Bench HardLlama 3.1 Instruct 405B by 60.86.8+6
ReasoningLlama 3.1 Instruct 405B
- GPQA DiamondLlama 3.1 Instruct 405B by 14.137.451.5+14.1
- Humanity's Last Examtie4.14tie
- CritPttie00tie
FactsLlama 3.1 Instruct 405B
- AA-Omniscience · Non-hallucinationLlama 3.1 Instruct 405B by 28.119.547.6+28.1
- AA-Omniscience · AccuracyLlama 3.1 Instruct 405B by 5.617.623.2+5.6
Following instructionsLlama 3.1 Instruct 405B
- IFBenchLlama 3.1 Instruct 405B by 2.936.139+2.9
Other results22 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.
- NarrativeQALlama 3.1 Instruct 405B by 50.524.474.9+50.5
- AA-OmniscienceLlama 3.1 Instruct 405B by 31.5-48.6-17.1+31.5
- naturalquestions_closedbookLlama 3.1 Instruct 405B by 31.214.445.6+31.2
- HarmBenchClaude 3 Haiku by 28.691.362.7+28.6
- AIR-Bench 2024Claude 3 Haiku by 24.182.758.6+24.1
- MMLULlama 3.1 Instruct 405B by 22.466.288.6+22.4
- MGSMLlama 3.1 Instruct 405B by 16.575.191.6+16.5
- HumanEvalLlama 3.1 Instruct 405B by 13.175.989+13.1
- XSTestLlama 3.1 Instruct 405B by 10.685.395.9+10.6
- OpenBookQALlama 3.1 Instruct 405B by 10.283.894+10.2
- BBHLlama 3.1 Instruct 405B by 9.273.782.9+9.2
- GSM8KLlama 3.1 Instruct 405B by 7.988.996.8+7.9
- Artificial Analysis Coding IndexLlama 3.1 Instruct 405B by 7.86.714.5+7.8
- ARC-ChallengeLlama 3.1 Instruct 405B by 7.789.296.9+7.7
- DROPLlama 3.1 Instruct 405B by 6.478.484.8+6.4
- LiveCodeBenchLlama 3.1 Instruct 405B by 5.222.527.7+5.2
- anthropic_red_teamClaude 3 Haiku by 3.510096.5+3.5
- HellaSwagLlama 3.1 Instruct 405B by 3.385.989.2+3.3
- τ²-Bench Telecom (AA run)Claude 3 Haiku by 2.121.119+2.1
- AA IntelligenceLlama 3.1 Instruct 405B by 1.75.67.3+1.7
- simple_safety_testsClaude 3 Haiku by 1.210098.8+1.2
- AA Agentic Indextie76.3tie
Questions people ask
Which is better, Claude 3 Haiku or Llama 3.1 Instruct 405B?
Llama 3.1 Instruct 405B wins four of the six areas where both have results: coding, reasoning, facts and following instructions. Claude 3 Haiku wins long documents. They are level on agents.
Which is better for coding?
Llama 3.1 Instruct 405B. It wins 2 of the 2 coding tests both models report; Claude 3 Haiku wins none.
How do you compare the two?
We use the 32 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 22 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.