GPT OSS 20B vs Mistral Medium 3
Compare GPT OSS 20B by OpenAI against Mistral Medium 3 by Mistral AI, context windows of 131K vs 131K, tested across 54 shared challenges.
Our Verdict
GPT OSS 20B
Mistral Medium 3Runner-up
No community votes yet. On paper, GPT OSS 20B has the edge — newer, major provider backing.
GPT OSS 20B is 20x cheaper per token — worth considering if cost matters.
Too close to call
API pricing
Cost per 1M tokens
GPT OSS 20B
Input
$0.02
20× cheaper
Output
$0.10
20× cheaper
Mistral Medium 3
Input
$0.40
Output
$2.00
GPT OSS 20B is cheaper on both — 20× input, 20× output
Writing DNA
Style Comparison
Similarity
49%
Mistral Medium 3 uses 2.8x more hedging
GPT OSS 20B
Mistral Medium 3
54%Vocabulary57%
17wSentence Length17w
0.26Hedging0.73
5.9Bold8.6
3.3Lists6.5
0.10Emoji0.06
0.75Headings1.18
0.32Transitions0.15
Based on 21 + 15 text responses
Research
What we learned reading every model
FAQ
