French AI startup Mistral has launched Mistral Medium 3, a new AI model designed to offer state-of-the-art (SOTA) performance at an 8X lower cost. Priced at $0.40 per million input tokens and $20.80 per million output tokens, the model claims to perform at or above 90% of the capabilities of Anthropic's Claude Sonnet 3.7, while also surpassing other recent models like Meta's Llama 4 Maverick and Cohere's Command A in AI performance evaluations. Mistral Medium 3 is tailored for coding and STEM tasks, offering high performance in these areas, and excels in multimodal understanding. It is deployable on any cloud, including self-hosted environments with four GPUs or more, and is available through Mistral's API, where enterprise customers can fine-tune it. The model is also accessible on Amazon's Sagemaker platform and will soon be available on Microsoft's Azure AI Foundry and Google's Vertex AI platforms. The company, founded in 2023, has raised over €1.1 billion and serves clients in financial services, energy, and healthcare. These sectors have been beta testing Mistral Medium 3 for applications such as customer service, workflow automation, and data analysis. Additionally, Mistral introduced Le Chat Enterprise, a customizable and secure AI assistant for businesses. Mistral teased the upcoming release of a larger model in the near future.