Post ContentMistral Saba is capable of generating the most accurate and relevant responses, even better than models that are five times its size. (Image: Mistral)
Mistral on Tuesday, March 17, announced a platform that lets businesses train custom AI models on their own enterprise data from scratch.
Mistral Forge lets customers build custom AI models using the French AI startup’s wide library of open-weight AI models, including small language models such as the recently unveiled Mistral Small 4. The Forge platform was announced on the sidelines of Nvidia GTC 2026, the chip giant’s annual technology conference that was held this year with a specific focus on agentic AI models for enterprises.
While customers can choose which model and infrastructure to use, Mistral said that companies looking for more guidance can bring in the company’s forward-deployed engineers who will work directly with the teams to use the right data and train AI models specific to their needs. Tech companies such as IBM and Palantir also follow a similar business model.
Mistral’s announcement reflects its pointed focus on building AI products for corporate clients even as US counterparts have pulled ahead in terms of consumer adoption. However, Mistral appears to be in the right lane as the focus among major AI players such as OpenAI is increasingly shifting toward enterprise solutions, driven by mounting pressure to show returns on AI investments.
Also Read | Mistral announces new AI model Medium 3 at 8x lower cost
With the Forge platform, Mistral is looking to bridge the gap by enabling enterprises to train models on company data from the start instead of providing access to fine-tuned models with techniques such as retrieval augmented generation (RAG). Mistral is on track to surpass $1 billion in annual recurring revenue this year, according to CEO Arthur Mensch.
In terms of availability, Mistral said that its Forge platform has already been made available to partners such as Ericsson, the European Space Agency, Italian consulting company Reply, and Singapore’s DSO and HTX. Dutch chipmaker ASML is also an early adopter of Forge.
The main use cases of the platform include developing models tailored to language and cultural needs for governments, models that adhere to high compliance requirements by financial firms, manufacturers with distinct customisation needs, software companies looking for AI models that can be tuned to their code base, etc.
© IE Online Media Services Pvt Ltd
