Databricks Invests in Mistral to Integrate AI Models into Data Intelligence Platform

Databricks, a rapidly growing data infrastructure company led by Ali Ghodsi, has announced a partnership with Mistral, a Paris-based startup renowned for its innovative large language models (LLMs), many of which are open-sourced.

This collaboration involves Databricks investing an undisclosed sum in Mistral, contributing to its Series A funding, and integrating select Mistral LLMs into its data intelligence platform. This integration simplifies the use of advanced generative AI models for enterprise users, maintaining the existing security, privacy, and governance protocols of Databricks.

Mistral has been proactive in forming strategic partnerships, recently aligning with Snowflake, a direct competitor in the data cloud sector, and Microsoft, a move that has attracted regulatory scrutiny.

As part of this new partnership, Databricks will natively integrate two text-generation models from Mistral: Mistral 7B and Mixtral 8x7B, both open-sourced. The Mistral 7B is a compact 7 billion parameter transformer model, while Mixtral 8x7B features a sparse mixture of experts (SMoE) with a context length of 32k. Notably, Mixtral 8x7B surpasses Meta’s Llama 2 70B and OpenAI’s GPT-3.5 in various benchmarks, enhancing its applicability.

Users of the Databricks Data Intelligence Platform can now access these models via the marketplace, complete with detailed capabilities and usage options. They can experiment with the models in the Mosaic AI Playground, utilize them as optimized endpoints through Mosaic AI Model Serving, or customize them with proprietary data via Mosaic AI Foundation Model Adaptation for specific use cases.

According to a Databricks spokesperson, while this partnership enhances generative AI capabilities for its users, there is currently no information regarding the integration of Mistral's latest model, Mistral Large. This model, which significantly outperforms competitors like Claude 2 and Google’s Gemini Pro, has been integrated into Snowflake’s Cortex service.

Mistral continues to expand its influence in the industry, having raised Europe’s largest-ever seed round in June 2023, followed by a substantial Series A. With significant investments—such as a recent $16 million from Microsoft to feature its models on the Azure platform—Mistral is solidifying its position as a trusted vendor in the AI landscape, competing with giants like OpenAI and Google. Partnerships with IBM, Perplexity, and Amazon further amplify its reach and application across various sectors.