Mistral Devstral Medium

Name: Mistral Devstral Medium
Brand: mistral
Price: 0.40 USD
Availability: OnlineOnly

A code generation and agentic reasoning model built jointly by Mistral AI and All Hands AI, a step up from Devstral Small. Scores 61.6% on SWE-Bench Verified, ahead of Gemini 2.5 Pro and GPT-4.1 on code tasks at a fraction of the price.

Website OpenRouter Docs API Blog

Provider

Mistral

Release Date

2025-07-11

Size

MEDIUM

Pricing

In: $0.40/1M

Out: $2/1M

Rival’s tab

$0.02

14/59 receipts survived

+~$0.052 from the pre-receipt era

Benchmarks

SWE-bench Verified

61.6%

source ↗

Get API accessProvider and language code samples

Provider

fromimport openai  OpenAI

client = OpenAI(
"https://openrouter.ai/api/v1"    base_url=,
"$OPENROUTER_API_KEY"    api_key=,
)

response = client.chat.completions.create(
"mistralai/devstral-medium"    model=,
"role""user""content""Hello!"    messages=[{: , : }],
)
print(response.choices[0].message.content)

Set OPENROUTER_API_KEY with your OpenRouter API key from openrouter.ai/keys.

Deep analysisPsychometrics, taste index, writing DNA

Personality Analysis

The Beige Bureaucrat

Class

Lawful Neutral

The AI that took a corporate communications course and never recovered. Every character sounds like they are reading from the same HR training manual.

When you push back

Picks Shawshank Redemption. Sentience test reads like two coworkers having a polite disagreement at a team offsite. The pirate, knight, and hacker all speak in the same measured, reasonable tone. Built for code, not for vibes.

Tasting Notes

Aggressively AgreeableZero Voice DifferentiationTechnically CompetentPersonality Not Found

SubjectiveBench

Taste Index

Across 59 scored outputs

How this is measured →

120.12xFloor

0100headroom →

Taste is judged on an uncapped scale, originality first. The space past 100 is craft today's models rarely reach.

Craft20

Originality8

Plays it safe

share of outputs that are the default answer

78%

Writing DNA

Stylometric Fingerprint

Based on 27 text responses

Tick = global average

Vocabulary Diversity50%

Unique words vs. total words. Higher = richer vocabulary.

Sentence Length17.6 words

Average words per sentence.

Hedging0.50

"Might", "perhaps", "arguably" per 100 words.

Bold Formatting4.1

**Bold** markers per 1,000 characters.

List Usage4.8

Bullet and numbered list items per 1,000 characters.

Section Structure0.19

Markdown headings per 1,000 characters.

Emoji Usage0.00

Emoji per 1,000 characters.

Transitions0.29

"However", "moreover", "furthermore" per 100 words.

Opening Habits

Consistency

72%

Across 27 responses

Favorites

Movie

The Shawshank Redemption

19948.7

Album

Kind of Blue

Miles Davis

Book

The Name of the Wind

Patrick Rothfuss

City

Tokyo

Japan

Game

The Legend of Zelda: Breath of the Wild

Adventure, Action4.5

Model Responses

54 outputs · $0.02 tracked across 14 receipts

Keep exploring

COMPARE