French AI startup Mistral launches AI model Medium 3
Mistral claims Medium 3 have outperformed rival models such as Claude Sonnet 3.7, Llama 4, Maverick, and GPT-4o on numerous benchmarks.
Paris-based AI startup Mistral has launched its latest AI model, Mistral Medium 3, built specifically for enterprise use.
According to a blog post, the company claims the model to have outperformed rival models such as Claude Sonnet 3.7, Llama 4, Maverick, and GPT-4o on numerous benchmarks.
The new model is 8x cost-effective and easier to deploy, helping businesses to adopt and scale AI solutions faster.
The AI model supports use cases such as coding and multimodal understanding, besides other enterprise capabilities, including hybrid or on-premises deployment, options for custom post-training, and integration into existing enterprise tools and systems.
“Mistral Medium 3 is designed to be frontier-class, particularly in categories of professional use. Performance accuracy on all benchmarks were obtained through the same internal evaluation pipeline. Mistral Medium 3 particular stands out in coding and STEM tasks where it comes close to its very large and much slower competitors,” the company said.
The company's applied AI solutions allow the model to be continuously pretrained, fully fine-tuned, and integrated into enterprise knowledge bases, making it well-suited for domain-specific training and adaptive workflows.
It added that early users in finance, energy, and healthcare sectors are utilising the model to improve customer service with deeper context.
Starting Thursday, the model can be accessed through API on its platform, La Plateforme, as well as on Amazon SageMaker. It will soon be available on other major cloud and AI platforms, including IBM WatsonX, NVIDIA NIM, Azure AI Foundry, and Google Cloud Vertex AI.
Edited by Suman Singh