Magistral Small 1

Magistral Small 1

Mistral · Released Jun 10, 2025 Professional
Intelligence #10 / 576
82.0 Our Score
AA Index #205 / 378
16.8 Artificial Analysis
Input
— Not priced
Output
— Not priced
Context
— Not reported

Analysis Summary

Magistral Small 1 is Mistral's compact reasoning model, released in June 2025. Despite its small size, it posts a GPQA of 0.641, livecodebench of 0.514, and MMLU-Pro of 0.746, which are competitive figures for its class. Its intelligence index of 16.8 and agentic index of 15.6 indicate meaningful reasoning and some agentic capability, though both remain well below frontier models.

For businesses, Magistral Small 1 suits structured content generation, code assistance, and lighter analytical tasks where cost efficiency matters. Its instruction-following score (ifbench 0.248) is reasonable, and its coding index of 11.1 limits it for heavy software engineering work. Agentic reliability is modest rather than production-grade for complex multi-step workflows.

As a Mistral model, it benefits from strong European enterprise adoption and flexible deployment options. Teams needing a capable, cost-effective reasoning model for mid-complexity tasks will find it a practical fit.

Assessed June 6, 2026

Editorial notes

Magistral Small 1 from Mistral delivers a strong GPQA of 0.641 and livecodebench of 0.514 for a small reasoning model, with agentic and instruction-following coverage, making it a capable lightweight option for structured business tasks.

Rankings consider pricing, capabilities, benchmarks, and real-world applicability and are refreshed as new models launch. Feedback?

Performance Profile

Intelligence3.2Technical2Value0Content2.8
Intelligence 3.2/10
Technical 2/10
Content 2.8/10
Value 0/10

How Magistral Small 1 compares

Magistral Small 1 ranks #205 of 378 AI models we track for overall intelligence, #223 of 315 for coding, #214 of 289 for agentic tasks. Magistral Small 1 is currently free to use via OpenRouter.

Performance Indices

Source: Artificial Analysis

16.8 Intelligence Index
11.1 Coding Index
15.6 Agentic Index
41.3 Math Index

Benchmark Scores

Intelligence

GPQA Diamond 64.1% Graduate-level scientific reasoning
HLE 7.2% Humanity's Last Exam
MMLU Pro 74.6% Multi-task language understanding
MATH 500 96.3% Mathematical problem-solving
AIME 71.3% Competition mathematics
AIME 2025 41.3% Competition mathematics (2025)
SciCode 24.1% Scientific computing

Technical

LiveCodeBench 51.4% Live coding evaluation
TerminalBench Hard 4.5% Agentic terminal tasks
τ²-Bench 26.6% Conversational agent benchmark

Content

IFBench 24.8% Instruction following

Benchmark data from Artificial Analysis and Hugging Face

How does Magistral Small 1 stack up?

Compare side-by-side with other professional models.

Compare Models

Model Information

ProviderMistral
Release Date June 10, 2025
Status Active

Leaderboard Categories

Frequently asked questions about Magistral Small 1

How much does Magistral Small 1 cost?

Magistral Small 1 is currently available for free via OpenRouter.

Is Magistral Small 1 good for coding?

On our coding benchmark index, Magistral Small 1 ranks #223 of 315 models, placing it in the broader range of the field for code generation and debugging.

Who created Magistral Small 1?

Magistral Small 1 is developed by Mistral and was released on June 10, 2025.