The Story

French AI startup Mistral has announced Forge, a new platform that enables enterprises to build custom AI models using their own proprietary data and infrastructure.

According to the official announcement from Mistral, Forge provides companies with the same training pipelines and infrastructure the company uses internally, allowing organizations to develop models without sharing sensitive data.

What Forge offers

• Full training pipeline: Forge supports pre-training, post-training, and reinforcement learning workflows similar to Mistral’s internal processes.
• Data privacy: Training can run entirely on a company’s own servers, ensuring zero data exposure to external systems.
• Enterprise-ready: Designed for industries like defense, finance, and government where data control is critical.

Early adoption and use cases

Mistral says early partners include ASML, Ericsson, and the European Space Agency, with applications ranging from legacy code migration to restoring historical manuscripts.

Part of a broader strategy

The launch comes alongside other recent releases from Mistral, including new models like Small 4 and Leanstral, as well as its participation in NVIDIA’s Nemotron ecosystem.

Why it matters

Most enterprise AI today relies on adapting general-purpose models through prompting or light fine-tuning. Forge represents a shift toward deeper customization, where companies can build models directly on their internal knowledge and systems.

This approach could be especially important for organizations dealing with sensitive data, strict compliance requirements, and domain-specific expertise.