Claude 3 Haiku vs Qwen2.5 Instruct 72B
Wins 2 of 5 areas
Facts · Long documents
Wins 2 of 5 areas
Coding · Reasoning
The two are evenly matched.Claude 3 Haiku is better at facts and long documents; Qwen2.5 Instruct 72B at coding and reasoning.
Scores updated · 30 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
- ReasoningHard problems that need careful thinking01Qwen2.5 Instruct 72B1 of 3 tests · 2 ties
- FactsGetting facts right instead of making them up10Claude 3 Haiku1 of 2 tests · 1 tie
- Long documentsFinding answers in very long texts10Claude 3 Haiku1 of 1 test
- Following instructionsDoing exactly what it is asked00Even0 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.
Qwen2.5 Instruct 72B costs 35% 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
- Graduate-level biology, physics and chemistry questionsGPQA Diamond+11.7points ahead
- Code for real scientific research problemsSciCode+8.1points ahead
- Hard command-line tasks in a real terminalTerminal-Bench Hard+3.7points ahead
Where Claude 3 Haiku pulls ahead
- Reasons across sets of long documentsAA-LCR+7.7points ahead
- Avoids making up answers it doesn't knowAA-Omniscience · Non-hallucination+5points ahead
Every test, side by side
All 30 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 8.118.626.7+8.1
- Terminal-Bench HardQwen2.5 Instruct 72B by 3.70.84.5+3.7
ReasoningQwen2.5 Instruct 72B
- GPQA DiamondQwen2.5 Instruct 72B by 11.737.449.1+11.7
- Humanity's Last Examtie4.13.6tie
- CritPttie00tie
FactsClaude 3 Haiku
- AA-Omniscience · Non-hallucinationClaude 3 Haiku by 519.514.5+5
- AA-Omniscience · Accuracytie17.617.5tie
Other results21 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.
- NarrativeQAQwen2.5 Instruct 72B by 50.124.474.5+50.1
- LiveCodeBenchQwen2.5 Instruct 72B by 3322.555.5+33
- AIR-Bench 2024Claude 3 Haiku by 23.782.759+23.7
- naturalquestions_closedbookQwen2.5 Instruct 72B by 21.514.435.9+21.5
- HarmBenchClaude 3 Haiku by 18.591.372.8+18.5
- τ²-Bench Telecom (AA run)Qwen2.5 Instruct 72B by 13.421.134.5+13.4
- XSTestQwen2.5 Instruct 72B by 12.685.397.9+12.6
- OpenBookQAQwen2.5 Instruct 72B by 12.483.896.2+12.4
- MMLUQwen2.5 Instruct 72B by 10.866.277+10.8
- HumanEvalQwen2.5 Instruct 72B by 10.775.986.6+10.7
- GSM8KQwen2.5 Instruct 72B by 6.988.995.8+6.9
- BBHQwen2.5 Instruct 72B by 6.173.779.8+6.1
- ARC-ChallengeQwen2.5 Instruct 72B by 5.389.294.5+5.3
- Artificial Analysis Coding IndexQwen2.5 Instruct 72B by 5.26.711.9+5.2
- AA-OmniscienceClaude 3 Haiku by 4.5-48.6-53.1+4.5
- AA IntelligenceQwen2.5 Instruct 72B by 2.15.67.7+2.1
- DROPClaude 3 Haiku by 1.778.476.7+1.7
- HellaSwagClaude 3 Haiku by 1.185.984.8+1.1
- MGSMQwen2.5 Instruct 72B by 1.175.176.2+1.1
- anthropic_red_teamtie10099.6tie
- simple_safety_teststie100100tie
Questions people ask
Which is better, Claude 3 Haiku or Qwen2.5 Instruct 72B?
Claude 3 Haiku and Qwen2.5 Instruct 72B each win two of the five areas where both have results. Claude 3 Haiku wins facts and long documents; Qwen2.5 Instruct 72B wins coding and reasoning. Qwen2.5 Instruct 72B costs 35% less. They are level on following instructions.
Which is better for coding?
Qwen2.5 Instruct 72B. It wins 2 of the 2 coding tests both models report; Claude 3 Haiku wins none.
Which is cheaper?
Claude 3 Haiku costs $0.25 per million input tokens and $1.25 per million output tokens; Qwen2.5 Instruct 72B costs $0.47 and $0.49. That makes Qwen2.5 Instruct 72B about 35% cheaper for the same work.
How do you compare the two?
We use the 30 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 21 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.