Gemini 2.5 Flash Preview (thinking) vs Kimi K3
Compare Gemini 2.5 Flash Preview (thinking) by Google AI against Kimi K3 by Moonshot AI, context windows of 1.0M vs 1.0M, tested across 20 shared challenges.
Our Verdict
Kimi K3
Gemini 2.5 Flash Preview (thinking)Runner-up
No community votes yet. On paper, Kimi K3 has the edge — bigger model tier, newer.
Gemini 2.5 Flash Preview (thinking) is 4.3x cheaper per token — worth considering if cost matters.
Too close to call
API pricing
Cost per 1M tokens
Gemini 2.5 Flash Preview (thinking)
Input
$0.17
17× cheaper
Output
$3.50
4.3× cheaper
Kimi K3
Input
$3.00
Output
$15.00
Gemini 2.5 Flash Preview (thinking) is cheaper on both — 17× input, 4.3× output
Writing DNA
Style Comparison
Similarity
52%
Kimi K3 uses 77.9x more emoji
Gemini 2.5 Flash Preview (thinking)
Kimi K3
53%Vocabulary58%
14wSentence Length21w
0.38Hedging0.46
4.4Bold3.6
4.1Lists3.4
0.00Emoji0.78
0.00Headings0.89
0.24Transitions0.17
Based on 7 + 27 text responses
Research
What we learned reading every model
FAQ
