Mistral Medium 3 Delivers Cost-Effective High-Performance AI Model for Coding and STEM

French AI startup Mistral has launched Mistral Medium 3, a new AI model that prioritizes efficiency without sacrificing performance. This model is accessible via Mistral’s API, priced competitively at $0.40 per million input tokens and $20.80 per million output tokens, making it an attractive option for enterprises seeking cost-effective AI solutions.

According to Mistral, Medium 3 performs at or above 90% of the capabilities of Anthropic’s more expensive Claude Sonnet 3.7 model across a wide range of benchmarks. It also outperforms recent open-source models such as Meta’s Llama 4 Maverick and Cohere’s Command A, establishing itself as a leading choice for AI-driven tasks.

Tokens, the fundamental units AI models process, are priced to offer significant value. One million tokens roughly equate to 750,000 words, which is about 163,000 words longer than the novel “War and Peace,” highlighting the model’s capacity to handle extensive data efficiently.

Mistral Medium 3 is versatile in deployment, supporting any cloud infrastructure as well as self-hosted environments with four or more GPUs. This flexibility enables organizations to integrate the model into existing systems seamlessly, optimizing both performance and cost.

Founded in 2023, Mistral is a frontier AI model lab focused on creating a suite of AI-powered services including chatbot platforms and mobile applications. The company has raised over €1.1 billion in funding and counts major clients like BNP Paribas, AXA, and Mirakl among its users.

Mistral Medium 3 excels particularly in coding and STEM-related tasks and demonstrates strong multimodal understanding capabilities. Beta testers from financial services, energy, and healthcare sectors are already applying the model for customer service automation, workflow optimization, and complex data analysis.

In addition to Mistral’s API, which allows enterprises to collaborate on fine-tuning the model, Mistral Medium 3 is now available on Amazon SageMaker. Upcoming availability on Microsoft Azure AI Foundry and Google Vertex AI platforms will further expand access and integration options.

This launch follows Mistral’s recent release of Mistral Small 3.1 and precedes the announcement of a larger model expected soon, signaling the company’s commitment to advancing AI capabilities with a focus on efficiency and accessibility.