Mistral AI Launches Magistral 1.2: New 24B Multimodal Reasoning Model
Mistral AI has launched Magistral 1.2, a 24-billion parameter lightweight model that combines advanced image recognition with structured chain-of-thought reasoning.
What happened?
Mistral AI has launched Magistral 1.2, a new 24-billion parameter lightweight model. Building on Mistral Small 3.2, the model has been trained using both supervised fine-tuning (SFT) and reinforcement learning (RL) to enhance its reasoning capabilities. In addition to improved language comprehension, Magistral 1.2 includes a proprietary visual encoder for multimodal image and text analysis, as well as an optimized structure for chain-of-thought reasoning.
Key facts
| Parameterstorlek | 24 miljarder (24B) |
|---|---|
| Basmodell | Mistral Small 3.2 |
| Noggrannhet europeiska språk | 92% |
| Träningsmetod | SFT och RL (förstärkningsinlärning) |
Why it matters
This launch marks a significant step for compact AI models capable of performing advanced logic without requiring massive computational resources. By mandating intermediate reasoning steps via specific markers, the model generates more explainable and accurate decisions, thereby increasing precision in complex tasks such as mathematics and logic puzzles.
Who is affected?
The model is aimed at AI developers, researchers, and enterprises looking to run efficient multimodal AI systems locally or in the cloud. It is particularly beneficial for applications in technical translation, education, game analysis, and automated problem-solving.
Impact on the EU
Magistral 1.2 features a strong emphasis on European languages, including English, French, German, and Spanish, achieving high accuracy in these contexts. The model is distributed in accordance with European guidelines for openness and data protection.
What else you should know
The use of special markers, such as [THINK] and [/THINK], allows developers to easily filter or review internal reasoning steps. Furthermore, integration with Unsloth and torchao facilitates local fine-tuning and quantization for companies with stringent data privacy requirements.
Quick answers about this story
Vad har hänt?
När hände det?
Varför spelar det roll?
Hur bra är språkstödet i EU?
The link opens in a new window and leads to the publisher's own site.
Källan har spårats automatiskt från utgivaren via Aheadlines signalkedja.
AI-verktyg i artikeln
Topics
Get similar news straight to your inbox
The reader's room
Send in a question or an addition. The newsroom reads everything before it's published and replies when relevant. No AI-generated text – just people.
Sign in to submit a comment or question.
Read the article through your role
- Decide whether this affects strategy over 6–12 months or is just noise.
- Discuss with leadership: do we own the right question or does ownership need to move?
- Ask: what risk are we taking by NOT acting on this this quarter?
Generated angle — not editorial analysis of "Mistral AI Launches Magistral 1.2: New 24B Multimodal Reason"