French AI startup Mistral has introduced Mistral Forge, a new platform designed to allow enterprises to build custom AI models trained on their own internal data. The announcement was made at Nvidia GTC, with the company positioning the offering as a response to limitations in models trained primarily on public internet data.
CEO Arthur Mensch indicated that Mistral’s enterprise-focused strategy is driving growth, with the company on track to exceed $1 billion in annual recurring revenue. Elisa Salamanca, Head of Product, explained that the platform enables organizations and governments to tailor AI systems to their specific operational needs.
Unlike approaches such as fine-tuning or retrieval-based methods, Mistral stated that Forge supports full model training, offering greater control, domain specificity, and adaptability. Timothée Lacroix, Co-founder and Chief Technologist, added that customization allows enterprises to optimize model performance for targeted use cases.
