Qwen: Qwen3 235B A22B Thinking 2507 vs Qwen: Qwen3.5 397B A17B
Compare Qwen: Qwen3 235B A22B Thinking 2507 and Qwen: Qwen3.5 397B A17B, both from Qwen, context windows of 131K vs 262K, tested across 53 shared challenges.
Our Verdict
Qwen: Qwen3.5 397B A17B
Qwen: Qwen3 235B A22B Thinking 2507Runner-up
No community votes yet. On paper, Qwen: Qwen3.5 397B A17B has the edge — newer, bigger context window.
Qwen: Qwen3 235B A22B Thinking 2507 is 6.0x cheaper per token — worth considering if cost matters.
Too close to call
API pricing
Cost per 1M tokens
Qwen: Qwen3 235B A22B Thinking 2507
Input
$0.11
5.5× cheaper
Output
$0.60
6.0× cheaper
Qwen: Qwen3.5 397B A17B
Input
$0.60
Output
$3.60
Qwen: Qwen3 235B A22B Thinking 2507 is cheaper on both — 5.5× input, 6.0× output
Writing DNA
Style Comparison
Similarity
57%
Qwen: Qwen3 235B A22B Thinking 2507 uses 2.4x more emoji
Qwen: Qwen3 235B A22B Thinking 2507
Qwen: Qwen3.5 397B A17B
56%Vocabulary54%
14wSentence Length14w
0.37Hedging0.39
5.7Bold6.9
4.0Lists4.2
0.54Emoji0.23
0.70Headings0.93
0.14Transitions0.17
Based on 22 + 23 text responses
Research
What we learned reading every model
FAQ
