Archive — US — techcrunch
Mistral releases a pair of AI reasoning models
Mistral, a French AI lab, has launched its first family of reasoning models called Magistral, including Small and Medium versions. These models aim to enhance consistency in tasks such as math and physics. Despite advances, they lag behind competitors like OpenAI and Google. Mistral claims Magistral increases speed and supports multiple languages, positioning itself for diverse enterprise applications.
Key points
- Mistral releases two AI reasoning models: Magistral Small (24 billion parameters) available for download, and Magistral Medium in preview on its platforms.
- Models designed for structured calculations, decision trees, and multi-step logic.
- Founded in 2023, Mistral has raised over €1.1 billion and aims to catch up with tech leaders in reasoning AI.
- Benchmarks show that Magistral Medium underperforms against competitors like Gemini 2.5 Pro and Claude Opus 4.
- Magistral claims to deliver answers 10 times faster than competitors and supports several languages.
- The release follows Mistral's debut of 'vibe coding' client and focus on enterprise-oriented tools.
Why it matters
- The growing importance of reasoning AI models reflects current trends in IT, emphasizing rapid, accurate data analysis for decision-making.
- Competitors like OpenAI and Google set high benchmarks for performance, influencing Mistral's strategic direction.
- The launch underlines a competitive landscape where AI firms are racing to enhance model efficiency and application diversity.
Mistral's launch of the Magistral models marks a significant step in its AI journey, addressing competitive pressures from established players while focusing on enterprise solutions and speed, yet still facing challenges in benchmark performance.