Mistral AI Unveils Forge Platform for Custom Enterprise AI Models

Mistral AI has introduced Forge, a platform that enables enterprises to build and train custom AI models grounded in their own proprietary knowledge, according to the company’s product page. The system is designed to help organizations create frontier-grade large language models that operate within their specific workflows, policies, and data environments.

Forge allows companies to train models across the full lifecycle — including pre-training, reinforcement learning, and evaluation — using internal documentation, codebases, and structured data. It supports dense and mixture-of-experts architectures, as well as multimodal inputs such as text, images, and audio. The platform also includes synthetic data generation and KPI-aligned evaluation frameworks to ensure that models meet enterprise standards.

Mistral AI says Forge gives organizations flexibility to deploy models across private cloud, on-premises, or Mistral-managed infrastructure. It includes governance and auditing features for compliance and security, along with lifecycle management tools for versioning, rollback, and drift detection.

Early partners include Ericsson, the European Space Agency, ASML, Reply, and Singapore’s DSO and HTX, which are using Forge to train models tailored to their operational and regulatory needs.