Models/Devstral Small (May '25)

MDL

Devstral Small (May '25)

Mistral

Verified against Artificial Analysis · Jul 13, 2026

Capability

percentile index

Price /M

$0 / $0

in / out

Context

—

Avg score

33.2

10 benchmarks

AA Intelligence

11.8

index

Weights

Closed

Released

May 21, 2025

About

Devstral Small (May '25) is a proprietary model in the Devstral family from Mistral. Benchmarked on 10 evals, averaging 33.2.

Benchmark scores · 10

avg 33.2 — higher is better

MATHartificial-analysis · 2026-07

68.4

MMLU-Proartificial-analysis · 2026-07

63.2

GPQA Diamondartificial-analysis · 2026-07

43.4

τ²-benchartificial-analysis · 2026-07

IFBenchartificial-analysis · 2026-07

31.6

AA Long-Context Reasoningartificial-analysis · 2026-07

26.7

LiveCodeBenchartificial-analysis · 2026-07

25.8

SciCodeartificial-analysis · 2026-07

24.5

Terminal-Bench Hardartificial-analysis · 2026-07

6.1

Humanity's Last Examartificial-analysis · 2026-07

The research behind this model

all papers mentioning it →

Inferring Code Correctness from SpecificationMay 28, 2026

Mentions matched by name in title/abstract — from the arXiv + HF daily corpus.

Details

PublisherMistral

FamilyDevstral

Parameters—

License—

WeightsClosed

ReleasedMay 21, 2025

History

full ledger →

Capability over time

Append-only ledger — every observed change to this model's numbers.

API providers

Artificial Analysis$0/M