Devstral Small vs Magistral Medium
Change modelsDevstral SmallMistral
Magistral MediumMistral
IndexAll categories, out of 100.
28.2 ±8.2
33.7 ±4.1 best
GapIndex points below the top.
−5.5 · no clear gap
Top
CoverageIndex weight that has results.
75%
95% best
SpeedOutput tokens per second.
—
—
EloLMArena rating and rank.
—
1255
ReleasedMonth of release.
2025-05
2025-06
Input / 1MUS dollars per 1M input tokens.
—
—
Output / 1MUS dollars per 1M output tokens.
—
—
1K chats2K tokens in, 500 out, each.
—
—
1K documents20K tokens in, 1K out, each.
—
—
1K agent tasks200K tokens in, 20K out, each.
—
—
ContextMaximum tokens in one request.
128K best
40K
AgenticMulti-step tasks with tools.
21.6
30.6 best
CodingCode writing and repair.
35.7
39.2 best
ReasoningLogic problems and puzzles.
28.5 best
28.4
MultimodalTasks with images and text.
—
39.2
KnowledgeFacts and expert knowledge.
24.2
35.2 best
MultilingualTasks in many languages.
—
—
InstructionTasks with strict prompt rules.
21.4 best
12.5
MathMath problems.
—
41.8
Benchmarks
7 sharedOnly Devstral Small: 1Only Magistral Medium: 24
Devstral Small
Magistral Medium