Mistral AI Announces Mistral Medium 3
Mistral AI has announced Mistral Medium 3, a new language model that the company says balances state-of-the-art performance with 8× lower costs and simpler enterprise deployability. The model is available today via Mistral La Plateforme and Amazon SageMaker, with additional platform support coming soon.
According to Mistral, the company has released Mistral Medium 3, a model designed to deliver frontier performance at what it claims is an order of magnitude lower cost than competitors. The company states that Mistral Medium 3 performs at or above 90 percent of Claude Sonnet 3.7 across benchmarks while priced at $0.40 per million input tokens and $2 per million output tokens.
Mistral says the model surpasses leading open models such as Llama 4 Maverick and enterprise models such as Cohere Command A, and beats cost leaders such as DeepSeek v3 on pricing. The model is positioned for professional use cases including coding, STEM tasks, and multimodal understanding, and can be deployed on-premises, in VPC environments, or on any cloud provider using four GPUs or more.
Enterprise features include custom post-training, continuous pretraining, and integration into enterprise knowledge bases. The API is available starting today on Mistral La Plateforme and Amazon SageMaker, with upcoming availability on IBM WatsonX, NVIDIA NIM, Azure AI Foundry, and Google Cloud Vertex. Mistral also indicated that a larger model is in development.
The Wire · Newsletter
One careful email,
every Monday.
The week's most important AI stories, lightly edited and personally vouched for. No autoplay, no spam, easy to leave.
Comments · 0
Sign in to join the discussion.
Be the first to leave a thought.