Magistral Small 1.2

Available

Mistral · 2025-09-17 · 32,000 tokens

An AI model from Mistral, strongest at reasoning, suited to a broad range of AI workloads.

Supported modalities:textcode

Quick Overview

Text Generation1/10

Code Generation1/10

Reasoning8/10

Multimodal1/10

Scores from leading benchmark suites.

artificial analysis intelligence11.5

artificial analysis coding14.7

artificial analysis math80.3

Latency and throughput performance.

—P50 Latency

144.224tokens/sec