French AI firm Mistral launched Mistral Forge at Nvidia’s GTC conference. The platform allows companies to build custom generative AI models from scratch using proprietary data. This move positions Mistral to compete with OpenAI and Anthropic by offering ground-up creation instead of fine-tuning.
The launch is part of a collaboration with Nvidia to co-develop a base model on Nvidia’s DGX Cloud. This foundation will support the upcoming Nvidia Nemotron 4 family of open models. The partnership provides enterprises an alternative to proprietary systems from Google and Microsoft-backed OpenAI to address data privacy needs.