IBM announced the availability of the Mistral Large model on Watsonx.ai, enabling enterprises to leverage a variety of high-quality, high-performance foundation models for customized deployments.

Watsonx.ai is IBM's enterprise AI studio for AI developers. In addition to IBM's Granite models, a variety of open source and commercial models are available. These can be modified for specific use cases or integrated into business solutions and applications.

IBM has already offered a small-scale model within Watsonx.ai, "Mixtral-8x7B," suitable for enterprise environments. This model can process 50% more data per second than existing models.

The newly added Mistral Large is offered as a subscription-based Software-as-a-Service (SaaS) model. The Mistral Large model handles complex enterprise-wide tasks requiring inference and multilingual capabilities, and features enhanced Augmented Search Generation (RAG) performance. This enables extended chat interactions and large document processing.

It can also be connected to external tools, such as user-defined functions and APIs. Its superior coding performance makes it easy to build applications tailored to specific needs. It has built-in 'guardrail' features that can be used as safety devices to build responsible AI.

Customers who deploy Mistral Large on the Watsonx platform can leverage additional enterprise-ready offerings, including data stores, prompt labs, model tuning, process monitoring, and governance capabilities. Customers can deploy models within Watsonx.ai in their preferred environment, whether on-premises or via a public cloud provider.

IBM offers a commercial version of Mistral AI on its Watson platform. IBM offers customer protection for Mistral Large through a limited intellectual property indemnification program. This marks the first time IBM has extended its customer warranty program for its AI model, the IBM Granite model, to a third-party foundation model.