Claude 3.7 Sonnet vs GPT-5
Compare Claude 3.7 Sonnet by Anthropic against GPT-5 by OpenAI, in 37 community votes, gpt-5 wins 75% of head-to-head duels, context windows of 200K vs 400K, tested across 52 shared challenges.
Reasoning: GPT-5 wins 83% of votes
Web Design: GPT-5 wins 56% of votes
Image Generation: GPT-5 wins 67% of votes
Conversation: GPT-5 wins 67% of votes
Analysis: GPT-5 wins 100% of votes
Our Verdict
GPT-5Winner
Claude 3.7 SonnetRunner-up
Pick GPT-5. In 37 blind votes, GPT-5 wins 75% of the time. That's not luck.
GPT-5 particularly excels in Analysis, Reasoning, Image Generation.
Clear winner
API pricing
Cost per 1M tokens
Claude 3.7 Sonnet
Input
$3.00
Output
$15.00
GPT-5
Input
$1.25
2.4× cheaper
Output
$10.00
1.5× cheaper
GPT-5 is cheaper on both — 2.4× input, 1.5× output
Writing DNA
Style Comparison
Similarity
44%
Claude 3.7 Sonnet uses 11.0x more headings
Claude 3.7 Sonnet
GPT-5
62%Vocabulary65%
35wSentence Length13w
0.99Hedging0.20
1.2Bold0.2
4.3Lists4.6
0.00Emoji0.00
1.78Headings0.16
0.23Transitions0.03
Based on 13 + 16 text responses
Research
What we learned reading every model
FAQ
